The Ultimate Comparison

[ahaslett]

Well I bought a used 300/4 for my OM1 rather than new Sony 200-600G for my A7Riv. The Sony gathers a wee bit more light and was a lot cheaper than a used 300/4 (a lot). However, the 300/4 is a lot lighter. More pixels on the subject vs subject detection, sort of a choice.

just as a choice between 200-"900"mm vs 600mm only

Amusing that someone compares a 200-600mm taped at 400mm vs a 50-200mm!

maybe it wasn't a direct comparison of lenses or lens choices, but of formats?

“I want to see if I can justify carrying this beast of a system around, when I can get the same results with something much smaller” he said in the introduction. I have a GM 100-400, which I bought instead of the smaller and cheaper PL 50-200m, and Sigma 100-400mm for a number of reasons. That or the Sigma 100-400 would be a more sensible comparator for the new OM tele zoom.

When comparing formats, you don’t limit yourself to a single lens but choose the one(s) that best fit your needs.

There are 100-400mm FF lenses, Sigma make one. I have the Sony (which is cheaper than the OM).

I could post a video with a 50-200mm taped at 50mm vs my Samyang 100mm T2.3 (all 280g of it), also the SY is just a bit cheaper, well actually £2,560 cheaper.

Indeed I should choose FF with my SY 100/2.2 as against buying an OM 50-200/2.8 to take headshots at 50mm. That would be a comparison between systems? Actually, I’d compare my Zuiko 50/2 macro against the SY and/or against my Sony macro.

I liked Danny’s comparisons more.

A

 

[SonyX]

Well I bought a used 300/4 for my OM1 rather than new Sony 200-600G for my A7Riv. The Sony gathers a wee bit more light and was a lot cheaper than a used 300/4 (a lot). However, the 300/4 is a lot lighter. More pixels on the subject vs subject detection, sort of a choice.

just as a choice between 200-"900"mm vs 600mm only

Amusing that someone compares a 200-600mm taped at 400mm vs a 50-200mm!

maybe it wasn't a direct comparison of lenses or lens choices, but of formats?

“I want to see if I can justify carrying this beast of a system around, when I can get the same results with something much smaller” he said in the introduction.

Yes, he should have stuck to one version of the story for the review. like comparing at 400mm only with "comparable" setups
"setup" is a combo of camera and lens and it's very strange when he call a 32mp camera and a taped 300-600mm lens "equivalent" to 20mp body and 100-400mm lens and comparing prices

I have a GM 100-400, which I bought instead of the smaller and cheaper PL 50-200m, and Sigma 100-400mm for a number of reasons. That or the Sigma 100-400 would be a more sensible comparator for the new OM tele zoom.

yes, sure

When comparing formats, you don’t limit yourself to a single lens but choose the one(s) that best fit your needs.

There are 100-400mm FF lenses, Sigma make one. I have the Sony (which is cheaper than the OM).

I could post a video with a 50-200mm taped at 50mm vs my Samyang 100mm T2.3 (all 280g of it), also the SY is just a bit cheaper, well actually £2,560 cheaper.

Indeed I should choose FF with my SY 100/2.2 as against buying an OM 50-200/2.8 to take headshots at 50mm. That would be a comparison between systems?

why not

"Well I bought a used 300/4 for my OM1 rather than new Sony 200-600G for my A7Riv."

For myself, I was choosing between 75/1.8 and 35-100/2.8, and stayed with a prime

Actually, I’d compare my Zuiko 50/2 macro against the SY and/or against my Sony macro.

I liked Danny’s comparisons more.

A

 

[tdekany]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

 

[unhappymeal]

Another day, another comparison against full frame thread started by a m4/3 user...

 

[ahaslett]

Well I bought a used 300/4 for my OM1 rather than new Sony 200-600G for my A7Riv. The Sony gathers a wee bit more light and was a lot cheaper than a used 300/4 (a lot). However, the 300/4 is a lot lighter. More pixels on the subject vs subject detection, sort of a choice.

just as a choice between 200-"900"mm vs 600mm only

Amusing that someone compares a 200-600mm taped at 400mm vs a 50-200mm!

maybe it wasn't a direct comparison of lenses or lens choices, but of formats?

“I want to see if I can justify carrying this beast of a system around, when I can get the same results with something much smaller” he said in the introduction.

Yes, he should have stuck to one version of the story for the review. like comparing at 400mm only with "comparable" setups
"setup" is a combo of camera and lens and it's very strange when he call a 32mp camera and a taped 300-600mm lens "equivalent" to 20mp body and 100-400mm lens and comparing prices

I have a GM 100-400, which I bought instead of the smaller and cheaper PL 50-200m, and Sigma 100-400mm for a number of reasons. That or the Sigma 100-400 would be a more sensible comparator for the new OM tele zoom.

yes, sure

When comparing formats, you don’t limit yourself to a single lens but choose the one(s) that best fit your needs.

There are 100-400mm FF lenses, Sigma make one. I have the Sony (which is cheaper than the OM).

I could post a video with a 50-200mm taped at 50mm vs my Samyang 100mm T2.3 (all 280g of it), also the SY is just a bit cheaper, well actually £2,560 cheaper.

Indeed I should choose FF with my SY 100/2.2 as against buying an OM 50-200/2.8 to take headshots at 50mm. That would be a comparison between systems?

why not

"Well I bought a used 300/4 for my OM1 rather than new Sony 200-600G for my A7Riv."

For myself, I was choosing between 75/1.8 and 35-100/2.8, and stayed with a prime

That makes sense. The SY with an APSC crop to an effective 26Mpix 75/1.9 (needs stopping down to f2.5) would be an interesting comparison to the mighty 75/1.8. It's lighter but not weather sealed. There is something filmic about the rendering of the SY that I can't pin down. At first I thought it was resolution based, but it beats the GM at 100mm f4.5, the 28-75/2.8 cropped, and the 20-70/4 cropped (all at f5.6).

View attachment b63163d97a6844c4a19e07a62327a6ce.jpg
Distant hare




Crop

I thought that wasn't too bad. Beat my wife's Swarowski binoculars anyway.

I have a mix of primes and zooms across 2 mounts, so I can choose a bag that fits my expected uses and environment. I was all set to buy a used 35-100/2.8 but then I realised that a discounted 40-150/2.8 suited me better (as part of a whole kit).

I must admit that I've wanted a 300/4 since they were announced but release was delayed. I don't think that was decisive but maybe it influenced how I shaped my Pugh matrix.

Actually, I’d compare my Zuiko 50/2 macro against the SY and/or against my Sony macro.

I liked Danny’s comparisons more.

A

A

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post

 

[Cafe Racer]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

--
Amateur photographers worry about equipment, professional photographers worry about money, masters worry about light

 

[tdekany]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

 

[Cafe Racer]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

I'm refering to m4/3rds when high ISO noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features, therefore not so useful in my experience.

 

[tdekany]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen. Anything under normal circumstances, the 3 AI programs that I have tried work fantastic. If there is so much noise in a photo, I would suggest faster lenses. Otherwise, we are back to - not using the right gear or settings. Which is not there fault of. these programs.

 

[tdekany]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing,

Amazing how many m43 fans have inept FF buddies

Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen. Anything under normal circumstances, the 3 AI programs that I have tried work fantastic. If there is so much noise in a photo, I would suggest faster lenses. Otherwise, we are back to - not using the right gear or settings. Which is not there fault of. these programs.

Despite the somewhat optimistic take on the efficacy of the various NR options . They are not a free lunch and obviously cannot create real detail that excess noise has wiped out. Either way it is a moot point as the software is usable on all formats

I think there are also very different opinions on how well NR works in different use cases those using very high ISO in decent light . Compared to those shooting moving subjects in murky low light

An example from DPreviews E-M1X samples

https://www.dpreview.com/sample-galleries/7114239063/olympus-e-m1x-sample-gallery/3244527197

NR ( DXO PR5 in this instance ) has indeed removed noise but there is no fine detail, and the software has attempted to artificially make up detail such as the branding on the jacket



--
Jim Stirling:
"To argue with a person who has renounced the use of reason, is like administering medicine to the dead." - Thomas Paine
Feel free to tinker with any photos I post

To be honest, I am much more concerned about producing anything but a snapshot. So for my usage, birds in flight in good to decent light, the LR AI produces excellent results. Also using an EM1x/em1 mark2. The OM 1 cameras have way better sensors.

--
Amateur photographers worry about equipment, professional photographers worry about money, masters worry about light

 

[ahaslett]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing,

Amazing how many m43 fans have inept FF buddies

Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen. Anything under normal circumstances, the 3 AI programs that I have tried work fantastic. If there is so much noise in a photo, I would suggest faster lenses. Otherwise, we are back to - not using the right gear or settings. Which is not there fault of. these programs.

Despite the somewhat optimistic take on the efficacy of the various NR options . They are not a free lunch and obviously cannot create real detail that excess noise has wiped out. Either way it is a moot point as the software is usable on all formats

I think there are also very different opinions on how well NR works in different use cases those using very high ISO in decent light . Compared to those shooting moving subjects in murky low light

An example from DPreviews E-M1X samples

https://www.dpreview.com/sample-galleries/7114239063/olympus-e-m1x-sample-gallery/3244527197

NR ( DXO PR5 in this instance ) has indeed removed noise but there is no fine detail, and the software has attempted to artificially make up detail such as the branding on the jacket

NR can provide a better looking image at the expense of losing details and inventing new ones. In that sense, it is just a broader and more sophisticated version of other image processing tools.

AI is more successful on images that fall well within the training sets, and where there is enough signal to make a robust attempt at removing noise and imagining detail - a bit like the way we all fill in the gaps in human speech in a noisy environment.

From my own tests, DeepPrime works better with lower res RAWs in highly noisy environments - less tendency to imagine artefacts from clumps of high-res noise. It also works better on faces than it does on landscapes (or indeed logos). Presumably it is capable of reconstructing logos that are different to the actual subject but are both plausible and visually convincing in the absence of other evidence.

In less noisy environments, AI does better with more resolution - more signal = more evidence to start from.

So AI can create a better looking image in a relatively narrow band of use cases around images that were already close to acceptable. Beyond that, you use smaller images or longer shutter speeds (tripod, IS) or shallower DoF (bigger lenses, ultimately bigger sensors).

I think your point Jim, is that AI expands the shooting envelope of all sensor sizes. Once the shooting envelope meets your needs, further expansion is an insurance policy, potentially an expensive one.

Arguing that something meets my needs and therefore should meet everyone's needs I'm sure has a name?

Andrew

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post

 

[Cafe Racer]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen.

OK then!

Anything under normal circumstances, the 3 AI programs that I have tried work fantastic.

Is there anything about people in a night street scene that are not positioned in the foreground that doesn't sound normal?

If there is so much noise in a photo, I would suggest faster lenses.

I would suggest you take some time to think about your statement above. The maximum aperture of a fast lens is of no benefit if f4 is required to achieve the appropriate depth of field. ISO will also have to be raised to achieve a shutter speed that prevents or limits motion blur in a night street scene. I can only assume lack of knowledge or tunnel vision is why this seems to get overlooked when people suggest a fast lens as the answer to high ISO noise.

Otherwise, we are back to - not using the right gear or settings.

M4/3rds; is that the wrong gear even though AI NR exists? Is there anything wrong with using an aperture of f4 for the appropriate depth of field and ISO 1600 to achieve a fast enough shutter speed to prevent motion blur in a night street scene?

Which is not there fault of these programs.

Like you said "IS there anyone out there not using AI NR, especially on high iso images?" Well I tried it but it was not very effective as described. I think it's fair to say that AI NR is not a total solution for high ISO noise if there's some tasks it can't handle.

 

[ahaslett]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing,

Amazing how many m43 fans have inept FF buddies

Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen. Anything under normal circumstances, the 3 AI programs that I have tried work fantastic. If there is so much noise in a photo, I would suggest faster lenses. Otherwise, we are back to - not using the right gear or settings. Which is not there fault of. these programs.

Despite the somewhat optimistic take on the efficacy of the various NR options . They are not a free lunch and obviously cannot create real detail that excess noise has wiped out. Either way it is a moot point as the software is usable on all formats

I think there are also very different opinions on how well NR works in different use cases those using very high ISO in decent light . Compared to those shooting moving subjects in murky low light

An example from DPreviews E-M1X samples

https://www.dpreview.com/sample-galleries/7114239063/olympus-e-m1x-sample-gallery/3244527197

NR ( DXO PR5 in this instance ) has indeed removed noise but there is no fine detail, and the software has attempted to artificially make up detail such as the branding on the jacket

NR can provide a better looking image at the expense of losing details and inventing new ones. In that sense, it is just a broader and more sophisticated version of other image processing tools.

AI is more successful on images that fall well within the training sets, and where there is enough signal to make a robust attempt at removing noise and imagining detail - a bit like the way we all fill in the gaps in human speech in a noisy environment.

From my own tests, DeepPrime works better with lower res RAWs in highly noisy environments - less tendency to imagine artefacts from clumps of high-res noise. It also works better on faces than it does on landscapes (or indeed logos). Presumably it is capable of reconstructing logos that are different to the actual subject but are both plausible and visually convincing in the absence of other evidence.

I actually find NR is more effective on higher res files ( from larger sensors ) than m43 as you typically have significantly more detail to work with. Assuming that I have no need to shoot them at equivalent settings which I never do

Either way the best approach whenever possible is to try and mitigate the noise at capture

Definitely works for baby’s eyelashes.

In less noisy environments, AI does better with more resolution - more signal = more evidence to start from.

So AI can create a better looking image in a relatively narrow band of use cases around images that were already close to acceptable. Beyond that, you use smaller images or longer shutter speeds (tripod, IS) or shallower DoF (bigger lenses, ultimately bigger sensors).

Better looking is often debatable exchanging noise for artificial detail and softness

I tend to prefer more light because DeepPrime isn’t that good at fractal stuff like foliage.

I think your point Jim, is that AI expands the shooting envelope of all sensor sizes. Once the shooting envelope meets your needs, further expansion is an insurance policy, potentially an expensive one.

Going down the more light gathering , shallower DOF etc road costs and size increases even within systems

Arguing that something meets my needs and therefore should meet everyone's needs I'm sure has a name?

Marriage ?

Andrew

I only got the extremes of your 3 examples.

Tripod or IS and more full well capacity are my preferred solution but people in artificial light where you need enough DoF really is my NR use case.

A

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post

 

[Cafe Racer]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen.

OK then!

Anything under normal circumstances, the 3 AI programs that I have tried work fantastic.

Is there anything about people in a night street scene that are not positioned in the foreground that doesn't sound normal?

If there is so much noise in a photo, I would suggest faster lenses.

I would suggest you take some time to think about your statement above. The maximum aperture of a fast lens is of no benefit if f4 is required to achieve the appropriate depth of field. ISO will also have to be raised to achieve a shutter speed that prevents or limits motion blur in a night street scene. I can only assume lack of knowledge or tunnel vision is why this seems to get overlooked when people suggest a fast lens is the answer to high ISO noise.

Otherwise, we are back to - not using the right gear or settings.

M4/3rds; is that the wrong gear even though AI NR exists? Is there anything wrong with using an aperture of f4 for the appropriate depth of field and ISO 1600 to achieve a fast enough shutter speed to prevent motion blur in a night street scene?

Which is not there fault of these programs.

Like you said "IS there anyone out there not using AI NR, especially on high iso images?" Well I tried it but it was not very effective as described. I think it's fair to say that AI NR is not a total solution for high ISO noise if there's some tasks it can't handle.

My minimal use of higher ISO's is for a similar scenario namely moving people in very poor light I think NR even the best of them is far from a panacea . The birders are shooting high ISO in decent light ( great light compared to our scenarios ) and see things differently .

They may be in for a shock if they try to shoot at night using a high ISO and fast shutter speed!

We often see high ISO shots posted from them suggesting settings such as 1/4000th at 12800 ISO is low light

--
Jim Stirling:
"To argue with a person who has renounced the use of reason, is like administering medicine to the dead." - Thomas Paine
Feel free to tinker with any photos I post

 

[tdekany]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen.

OK then!

Anything under normal circumstances, the 3 AI programs that I have tried work fantastic.

Is there anything about people in a night street scene that are not positioned in the foreground that doesn't sound normal?

If there is so much noise in a photo, I would suggest faster lenses.

I would suggest you take some time to think about your statement above. The maximum aperture of a fast lens is of no benefit if f4 is required to achieve the appropriate depth of field. ISO will also have to be raised to achieve a shutter speed that prevents or limits motion blur in a night street scene. I can only assume lack of knowledge or tunnel vision is why this seems to get overlooked when people suggest a fast lens as the answer to high ISO noise.

Otherwise, we are back to - not using the right gear or settings.

M4/3rds; is that the wrong gear even though AI NR exists? Is there anything wrong with using an aperture of f4 for the appropriate depth of field and ISO 1600 to achieve a fast enough shutter speed to prevent motion blur in a night street scene?

Which is not there fault of these programs.

Like you said "IS there anyone out there not using AI NR, especially on high iso images?" Well I tried it but it was not very effective as described. I think it's fair to say that AI NR is not a total solution for high ISO noise if there's some tasks it can't handle.

I didn't single out m4/3 specifically. Every format can and does benefit from these AI programs. Now, your specific case, at night, bunch of people on a dark street? How do you even know that you need f4? But you have to pick your poison unfortunately. DOF, (at ight?), Noise? Something has to give, but your specific scenario is very are. Get a larger sensor camera maybe, or simply don't take the shot. You should, as a 'photographer" know the limits. Or, use your latest smart phone. But I admit defeat, you win, because using your logic and your scenario, AI noise reduction is not there yet, as you say. I certainly have waisted enough time of my life trying to be reasonable. Time I will never get back. So let me instead, use my time to take a Ferrari off road and complain that for all this money, it can't rock crawl. LOL! Enjoy your evening.

 

[tdekany]

So let me instead, use my time to take a Ferrari off road and complain that for all this money, it can't rock crawl. LOL! Enjoy your evening.

Now AI can solve that little issue

That really is funny. Thank you. See if AI can also add front and rear lockers.

--
Amateur photographers worry about equipment, professional photographers worry about money, masters worry about light

 

[bobnn]

I can't wait until we get to XenForo, so that all these huge unedited quotes will be dealt with.

 

[searun]

How about using an equivalent aperture? He compared f/4 to f/2.8. It matters.
--------------
Searun

 

[Cafe Racer]

Seems fair.

Back in my Pentax daze, I had both a 400/5.6 and 400/4, with the latter lens being twice the weight of the former. Some days I really want my "best", other days I wanted more convenience.

If we use that comparison logic, then the review should have been the FF 300-600mm vs MFT 150-400mm as the 50-200mm is not the best choice for wildlife while the FF 300-600mm is.

It is fair enough, all photos were taken at the same focal length etc, I don't think it is sponsored by OM Systems and the feller seems like a good genuine person as well

I enjoyed the video, but I have 2 observations. The noise in high iso shots are a non issue anymore, thanks to AI NR.

Not related to wildlife but when I've tried out AI NR (Topaz & DxO) on m4/3 images, I found they could not restore facial detail ruined by noise at high ISO (night shots), where the faces were not in the foreground. So for this reason I concluded that AI NR did not represent a total solution. Do you know if these AI NR software programs have been improved to make this now a non issue?

Like why not show the results afterwards? IS there anyone out there not using AI NR, especially on high iso images? Anyway. The other "mistake" was to not use the Sony 200-600 instead. Of course the 3rd party lens isn't going to focus as well as a native lens.

When was the last time you have tried it?

Last year. Have these AI NR software programs been improved to make this now a non issue?

Just the other day, my buddy, who shoots a Nikon Z8 shot in a very dark club. So his iso was at 16000. He is pretty new to photography and especially post processing, Uses LR. At the default settings, plus what ever else he did trying to improve the files, made the skin on the people too smooth. but once he made some more adjustments and reduces some settings, the finish looked great. But if you looked at the photos at 200%, you could see some noise. So you be the judge. He only has the 3 2.8 zooms and he did the best he could. 3 shots will be printed in 36x24 for the band leaders. if you want to find imperfections like gear heads usually do, you will find something. But I tell you what, not having AI, those photos would have been in the trash. If you just looked at these photos on the screen a normal viewing distances, you'd love them. You would never know or guess that they were shot at those ISOs. Hope this helps.

What I'm talking about is when noise has destroyed facial features. Faces that are not in the foreground are more susceptible to facial features being lost to noise in the circumstance I described. AI NR could remove noise but could not reconstruct the facial features.

Listen.

OK then!

Anything under normal circumstances, the 3 AI programs that I have tried work fantastic.

Is there anything about people in a night street scene that are not positioned in the foreground that doesn't sound normal?

If there is so much noise in a photo, I would suggest faster lenses.

I would suggest you take some time to think about your statement above. The maximum aperture of a fast lens is of no benefit if f4 is required to achieve the appropriate depth of field. ISO will also have to be raised to achieve a shutter speed that prevents or limits motion blur in a night street scene. I can only assume lack of knowledge or tunnel vision is why this seems to get overlooked when people suggest a fast lens as the answer to high ISO noise.

Otherwise, we are back to - not using the right gear or settings.

M4/3rds; is that the wrong gear even though AI NR exists? Is there anything wrong with using an aperture of f4 for the appropriate depth of field and ISO 1600 to achieve a fast enough shutter speed to prevent motion blur in a night street scene?

Which is not there fault of these programs.

Like you said "IS there anyone out there not using AI NR, especially on high iso images?" Well I tried it but it was not very effective as described. I think it's fair to say that AI NR is not a total solution for high ISO noise if there's some tasks it can't handle.

I didn't single out m4/3 specifically. Every format can and does benefit from these AI programs. Now, your specific case, at night, bunch of people on a dark street?

On the streets of London after dark there's light being emitted from various sources including street lights.

How do you even know that you need f4?

How do you know what aperture you need when you are shooting? Is it arbitrary or is it based on knowledge you've acquired?

But you have to pick your poison unfortunately. DOF, (at ight?), Noise? Something has to give, but your specific scenario is very are.

No the scenario is not very rare. Are you aware that in the winter in the UK it starts getting dark from between 3.45pm to 4.30pm? Do you think that no-one in the UK takes photos on the streets after 4.30pm in the winter?

Get a larger sensor camera maybe, or simply don't take the shot.

Why, because AI NR software doesn't represent a total solution for high ISO noise reduction?

You should, as a 'photographer" know the limits.

You should, as a 'photographer' do some more homework!

Or, use your latest smart phone. But I admit defeat, you win,

You admit you 'lose' the defensive argument you initiated.

because using your logic and your scenario, AI noise reduction is not there yet, as you say.

Try reading the title of my post: "AI NR not a total solution."

I certainly have waisted enough time of my life trying to be reasonable.

Defensive.

Time I will never get back. So let me instead, use my time to take a Ferrari off road and complain that for all this money, it can't rock crawl. LOL! Enjoy your evening.

Oh boy!

 

[the bluesman]

Another day, another comparison against full frame thread started by a m4/3 user...

Huh, hardly and I didn't make the video I just shared it as it is interesting, I don't and never start those kind of threads