A77 vs A37 noise/iq@3200

Started Jul 28, 2012 | Discussions
brian14478
Senior MemberPosts: 3,397
Like?
A77 vs A37 noise/iq@3200
Jul 28, 2012

these were screenshots from imaging resource comparometer..not meant to bash equipment. i got burned with reviews and my own ideas about an a33 being able to at least shoot 800iso with descent noise control.had i checked on imaging resource first i wouldnt have purchased the camera(it was returned)i am looking to get a a 57 or a37 and just figured people would like to see some unbiased test on camera bodies. the a77 im sure is a great camera i have seen some great pictures with it,but ive seen some great pictures from every alpha from the a100 up.if your looking into a newer body this site is an excellent reference tool..brian a37 on right

Jigal
Contributing MemberPosts: 547Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to brian14478, Jul 28, 2012

Exposure...

ISO settings can't be compared between cameras. See the effect of exposure on detail and noise in high ISO shots.

  • extra fine JPG

  • low NR

  • contrast -2

  • sharpen -2

  • saturation +0

-f/6.3

  • 1/40, 1/30, 1/25, 1/20, 1/15, 1/13 sec.

The +1.0eV shot has some motion blur
Focus was locked between shots, only exposure compensation changed.

Details and noise are quite different.

-- hide signature --

Regards,

Jigal.

 Jigal's gear list:Jigal's gear list
Olympus E-330 Sony SLT-A77 Olympus Zuiko Digital ED 18-180mm 1:3.5-6.3 Tamron 18-270mm F/3.5-6.3 Di II PZD +2 more
Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to brian14478, Jul 28, 2012

brian14478 wrote:

these were screenshots from imaging resource comparometer.

And they tell us that when you give cameras different amount of light, as they did, you get different amounts of noise. Hardly a startling revelation, and says nothing at all about the equipment involved.

just figured people would like to see some unbiased test

Lack of bias doesn't help when the test is broken.

These images tell us nothing at all about the camera bodies in question.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
brian14478
Senior MemberPosts: 3,397
Like?
Re: A77 vs A37 noise/iq@3200
In reply to theswede, Jul 28, 2012

theswede wrote:

brian14478 wrote:

these were screenshots from imaging resource comparometer.

And they tell us that when you give cameras different amount of light, as they did, you get different amounts of noise. Hardly a startling revelation, and says nothing at all about the equipment involved.

just figured people would like to see some unbiased test

Lack of bias doesn't help when the test is broken.

These images tell us nothing at all about the camera bodies in question.

Jesper

they were both still life shots at 3200 iso..didnt look at the others compared too much but i did pair up other bodies with a37 and a57,didnt try to say hey look at the comparison between sony and canon..i think its a sign of sony moving in a positive direction..if you look at a33 and others now discontinued compared to any of the offerings still in production you will see acceptable noise control/detail ect. all across the board(from iso up to 1280)including the a77...to everyone it should be good news with the a37 being the lowest priced entry level to date having the same sensor as the a57..i personally wouldn't even care about 3200iso too much it was just an extreme example of the least expensive sony yet.brian to me its great news cause i'll be getting another body soon though i love my a100 ..it should be called my a100iso

Reply   Reply with quote   Complain
thubleau7
Contributing MemberPosts: 516
Like?
Errr,,,,,,,,,,,,,what do you mean ?
In reply to theswede, Jul 28, 2012

According to the test that DPR did with the A57 the A65 and the A77 they were all done with the same lens and all at f/8 with manual white balance and default parameters.
I take it the A37 would be the same sensor as the A57 ?

So, where is there any difference in the light ? have you got the figures on the shutter speed on the DPR site ?

Reply   Reply with quote   Complain
brian14478
Senior MemberPosts: 3,397
Like?
Re: Errr,,,,,,,,,,,,,what do you mean ?
In reply to thubleau7, Jul 28, 2012

thubleau7 wrote:

According to the test that DPR did with the A57 the A65 and the A77 they were all done with the same lens and all at f/8 with manual white balance and default parameters.
I take it the A37 would be the same sensor as the A57 ?

So, where is there any difference in the light ? have you got the figures on the shutter speed on the DPR site ?

..i dont know what he was talking about though ive heard about the same reasoning before..imaging resource still life comparison at iso 100-6400 side by side...its interesting to look back at older models you may have owned...after the a33 i was considering a another ccd alpha with more mps and lcd view ect..but at400iso^things started getting ugly...brian

Reply   Reply with quote   Complain
altendky
Contributing MemberPosts: 613
Like?
Re: Errr,,,,,,,,,,,,,what do you mean ?
In reply to brian14478, Jul 28, 2012

brian14478 wrote:

..i dont know what he was talking about though ive heard about the same reasoning before..

I believe that DPR specifically adjusts light levels so that all cameras result in the same image levels for a given gray card at some exposure setting. So, they intentional change the scene per camera... but you were talking about imaging resource. If we check EXIF on the two images you referenced, they both match with Manual exposure, 1/640 sec, f/8, ISO 3200 with both using the Minolta AF 100-300mm F4.5-5.6 APO (D) or Sigma Lens Shot at 70 mm. On the other hand, at DPR:

A77
Lens: Sony DT 50mm F1.8 SAM (SAL50F18) or Tamron Lens
Exposure: Manual exposure, 1/1,000 sec, f/8, ISO 3200

A35
Lens: Sony 50mm F1.4 (SAL50F14)
Exposure: Manual exposure, 1/800 sec, f/8, ISO 3200

A57
Lens: Sony DT 50mm F1.8 SAM (SAL50F18) or Tamron Lens
Exposure: Manual exposure, 1/800 sec, f/8, ISO 3200

Yes, I know that these are not the cameras or photos you were talking about, but I believe it is what people are concerned about when only test images are provided without clarifying that the exposures were matched.

Oh, and don't forget focus when considering detail. :]

Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to brian14478, Jul 28, 2012

brian14478 wrote:

they were both still life shots at 3200 iso

No, they were not. They were both shot with the camera set at 3200 ISO, but both cameras did not use 3200 ISO. The amount of light for each camera is different. DPR and IR both do this, and admit this.

Thus, the comparison is useless. It tells you nothing at all about the relative performance of the cameras.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: Errr,,,,,,,,,,,,,what do you mean ?
In reply to thubleau7, Jul 28, 2012

thubleau7 wrote:

According to the test that DPR did with the A57 the A65 and the A77 they were all done with the same lens and all at f/8 with manual white balance and default parameters.

Which for starters is flawed. Which experienced user will have the camera at default parameters when stressing it?

I take it the A37 would be the same sensor as the A57 ?

Yes.

So, where is there any difference in the light ? have you got the figures on the shutter speed on the DPR site ?

DPR lists their exposure. It varies by huge amounts between cameras.

The whole comparison thing is worse than useless; it's actively misleading and should be deleted completely, or redone properly.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
Jigal
Contributing MemberPosts: 547Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to theswede, Jul 28, 2012

theswede wrote:

brian14478 wrote:

they were both still life shots at 3200 iso

No, they were not. They were both shot with the camera set at 3200 ISO, but both cameras did not use 3200 ISO. The amount of light for each camera is different. DPR and IR both do this, and admit this.

Thus, the comparison is useless. It tells you nothing at all about the relative performance of the cameras.

See the image in my reply in this thread. It is made of 100% crops of six different exposures with the camera at 3200 ISO. Noise levels and amount of detail varies a lot as you can see.

I hope this illustrates the argument of theswede a bit.

-- hide signature --

Regards,

Jigal.

 Jigal's gear list:Jigal's gear list
Olympus E-330 Sony SLT-A77 Olympus Zuiko Digital ED 18-180mm 1:3.5-6.3 Tamron 18-270mm F/3.5-6.3 Di II PZD +2 more
Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to Jigal, Jul 28, 2012

Jigal wrote:

I hope this illustrates the argument of theswede a bit.

Thank you for the images. They illustrate part of the point; that exposure level will in itself cause variation in the noise. If the tester decides that an image "looks good" while it is exposed lower than another testers image, it will look noisier.

But in addition there is a major complication. Camera ISO varies between manufacturers and models. Thus, even if the images look identically exposed and have similar noise levels, the actual light which has reached them can differ by a large amount. And it's not controversial to say that the A77 and A65 underestimate their ISO by a fair bit.

Thus, the 3200 from the A77 is much higher than the 3200 from the A37, as can be seen in exposure - and on top of this variation in exposure which may not cause much visible difference otherwise will cause big differences in noise.

There are simply too many variables to make tests like these worth anything at all. An experienced user of a camera will know what settings to use to get best results, and will expose to the right as much as the camera can handle, and the result will look completely different from these test images. Perversely, the further from the ideal the camera's default settings are and the more understated its exposure and actual ISO, the worse it will do in tests - and the better it will do in practice.

Thus, these tests are an abomination and need to go away or be corrected. They make things worse, not better, when deciding on cameras for low light use.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
chlamchowder
Senior MemberPosts: 1,982Gear list
Like?
If you want ISO performance
In reply to brian14478, Jul 28, 2012

If ISO performance is a primary concern, the best Alpha camera available for that is the a580. The a580 should be more than a stop better than the a100, and about 1/3-1/2th of a stop better than any of the currently available SLT models.

DxOMark compensates for differences in resolution, and presents data in a way that prevents exposure differences from having an impact on the scores. The downside is that you don't actually get to see sample images, but I think DxO still provides a good starting point for comparing cameras. DxO places the a580's ISO performance ahead of every Sony APS-C camera.

Anyways, here's an ISO 1600 image from the a580 (converted from raw). The noise is very light, even in the dark areas.

However, the JPG engine isn't as good. At high ISOs (like 6400), it likes to preserve color accuracy and saturation at the expense of visible color noise.

At ISO 12800, the images are still usable with some processing from the raw converter:

Processed with DxO:

 chlamchowder's gear list:chlamchowder's gear list
Sony Alpha DSLR-A580 Nikon D600 Sony DT 18-55mm F3.5-5.6 SAM +8 more
Reply   Reply with quote   Complain
altendky
Contributing MemberPosts: 613
Like?
Re: A77 vs A37 noise/iq@3200
In reply to theswede, Jul 28, 2012

theswede wrote:

brian14478 wrote:

they were both still life shots at 3200 iso

No, they were not. They were both shot with the camera set at 3200 ISO, but both cameras did not use 3200 ISO. The amount of light for each camera is different. DPR and IR both do this, and admit this.

I remember reading the DPR statement (I think), where does IR state this? In this particular case both cameras used the same lens (model at least) and were in manual with the same settings. Certainly they still could have changed light levels and the PP'ed back to the same luminosity, but where is this said?

I'm certainly not saying you're wrong but we are supposed to get references, yes? :]

Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to altendky, Jul 28, 2012

altendky wrote:

I remember reading the DPR statement (I think), where does IR state this?

Actually they didn't; I recalled wrong. They state they use gels to correct color temperature, but they do not mention how they get the exposure the same. At least not anywhere I find. But since they do, they normalize either input light or in PP.

In this particular case both cameras used the same lens (model at least) and were in manual with the same settings.

Which means they should have very different visible light levels, since they have very different actual ISO, according to pretty much any source of such metering. Look at dxomark for example.

There is no way to get the images looking identical with identical exposure and identical ISO setting at identical available light. Thus, they have changed something. Whatever it is, it's rendered the comparison a non-comparison.

I'm certainly not saying you're wrong but we are supposed to get references, yes? :]

Or use our heads. Actual ISO between cameras varies due to a myriad of reasons, so it doesn't make sense to assume that using the same setting between cameras is enough to ensure the actual ISO is equivalent.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
altendky
Contributing MemberPosts: 613
Like?
Re: A77 vs A37 noise/iq@3200
In reply to theswede, Jul 28, 2012

theswede wrote:

altendky wrote:

I remember reading the DPR statement (I think), where does IR state this?

Actually they didn't; I recalled wrong. They state they use gels to correct color temperature, but they do not mention how they get the exposure the same. At least not anywhere I find. But since they do, they normalize either input light

Which would ruin the comparison...

or in PP.

Which is what we would want, yes?

In this particular case both cameras used the same lens (model at least) and were in manual with the same settings.

Which means they should have very different visible light levels, since they have very different actual ISO, according to pretty much any source of such metering. Look at dxomark for example.

The A37 doesn't seem to have any data on DXO.

There is no way to get the images looking identical with identical exposure and identical ISO setting at identical available light. Thus, they have changed something. Whatever it is, it's rendered the comparison a non-comparison.

Isn't PP normalization appropriate?

I'm certainly not saying you're wrong but we are supposed to get references, yes? :]

Or use our heads. Actual ISO between cameras varies due to a myriad of reasons, so it doesn't make sense to assume that using the same setting between cameras is enough to ensure the actual ISO is equivalent.

I'm not sure why we care about " actual ISO ". Don't we care about having the same shutter speed (as we often choose that to avoid motion blur etc) and aperture (for depth of field control or sharpness) and then adjusting the exposure and resolution (we can leave up vs. downsampling to other threads) to match? We then had the same input scenario (lighting, subject, and physical light exposure settings) and have 'matching' output. Anyways, it has been shown here that ISO vs. PP doesn't seem to make a difference.

Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to altendky, Jul 28, 2012

altendky wrote:

or in PP.

Which is what we would want, yes?

Absolutely not!

The A37 doesn't seem to have any data on DXO.

Look at the A57. Same sensor, microlense arrangement and processor, so the values will be if not identical then at least quite similar.

Isn't PP normalization appropriate?

No. No normalization is appropriate. What is appropriate is adjusting the cameras actual ISO until the histograms match. Anything else is at best a waste of time but more usually misleading and harmful to our understanding of the cameras actual performance.

I'm not sure why we care about " actual ISO ".

Because that is what we will actually use when we take our photographs in low light. For example, if the A77 needs indicated ISO 2000 to take a certain photo, and the A37 needs indicated ISO 3200 to take the exact same photo, what is the value in having a comparison where both cameras were set to ISO 3200 and the images normalized? It will not show us what will happen in reality.

When we take photos in the real world we can't keep adjusting light levels until the ISO we want can be selected. If we could, why would we even need high ISO? Therefore what is interesting is how the cameras perform when light and exposure are identical - which means the ISO settings will be different among the cameras.

Don't we care about having the same shutter speed (as we often choose that to avoid motion blur etc) and aperture (for depth of field control or sharpness) and then adjusting the exposure and resolution (we can leave up vs. downsampling to other threads) to match?

You're skipping one step. You're saying "adjusting the exposure", after already fixing shutter and aperture. What exactly are you suggesting should be adjusted?

What should be adjusted is the camera ISO, until the histograms match. Anything else is wrong.

We then had the same input scenario (lighting, subject, and physical light exposure settings) and have 'matching' output.

If your magic step can be implemented we do. Since it is not implemented in these tests, the testers change either the input or the output to normalize, and either of them will destroy any ability to compare images.

Anyways, it has been shown here that ISO vs. PP doesn't seem to make a difference.

It makes a huge difference. Underexposing and pushing in PP will cause horrid noise. Much better to ETTR. This is one of the other large problems with these comparisons; at the very least the histogram should be shown so it can be compared.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
EvilOne
Forum ProPosts: 13,537Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to theswede, Jul 28, 2012

Noise seems to get so much attention here because it seems to be the weakest link in the design... Ive seen countless post and threads about this issue, and while it is an issue for sure.. it doesn't seem to effect my personal photographic style... I know its there, I can deal with it in PP. But even with it there is like having character.. its not like its a flesh eating virus...its some noise... I have several Microsoft screen savers and back grounds, offered by one of the most affluent companies on the planet and those images are so filled with noise and CA, yet millions and millions of Microsoft Windows 7 users have them on their desk top.. Just go into desk top themes and look at cats.. ( specifically black cats ) or look at rice patties..

While I guess it is something to bitch about. ( this noise issue ) it takes up alot of time for so many people who claim to take photos.
--
Bill aka EO

Reply   Reply with quote   Complain
altendky
Contributing MemberPosts: 613
Like?
Re: A77 vs A37 noise/iq@3200
In reply to theswede, Jul 28, 2012

theswede wrote:

Don't we care about having the same shutter speed (as we often choose that to avoid motion blur etc) and aperture (for depth of field control or sharpness) and then adjusting the exposure and resolution (we can leave up vs. downsampling to other threads) to match?

You're skipping one step. You're saying "adjusting the exposure", after already fixing shutter and aperture. What exactly are you suggesting should be adjusted?

That was a misuse of terminology, I should have said 'adjusting luminosity and resolution'.

Anyways, it has been shown here that ISO vs. PP doesn't seem to make a difference.

It makes a huge difference. Underexposing and pushing in PP will cause horrid noise. Much better to ETTR. This is one of the other large problems with these comparisons; at the very least the histogram should be shown so it can be compared.

Yes, I misused exposure above, but not here. I didn't say underexpose and PP, I said ISO vs. PP. Here is what I was referring to:

http://forums.dpreview.com/forums/read.asp?forum=1037&message=38125800

For a given exposure (shutter and aperture) it doesn't seem to matter if you bump ISO or PP luminosity. At least that's what I get from the comparisons. And that's my real point. Stop worrying about the ISO as it is just a tool for luminosity adjustment same as PP. Of course, that sample was only from one camera, perhaps others are different or maybe there was a flaw in the test?

Reply   Reply with quote   Complain
shady1991
Regular MemberPosts: 207Gear list
Like?
Re: If you want ISO performance
In reply to chlamchowder, Jul 28, 2012

Are those pictures taken with beercan?
P.S. excellent shots!

 shady1991's gear list:shady1991's gear list
Sony SLT-A57
Reply   Reply with quote   Complain
theswede
Veteran MemberPosts: 3,936Gear list
Like?
Re: A77 vs A37 noise/iq@3200
In reply to altendky, Jul 28, 2012

altendky wrote:

That was a misuse of terminology, I should have said 'adjusting luminosity and resolution'.

Do you mean changing the amount of light in the scene? How do you propose that results in anything which can be compared?

Yes, I misused exposure above, but not here. I didn't say underexpose and PP, I said ISO vs. PP.

That is exactly what it is; underexpose and PP.

For a given exposure (shutter and aperture) it doesn't seem to matter if you bump ISO or PP luminosity.

You can't do it in jpeg. Thus, this is not what anyone does, since all shots are at default jpeg settings.

Stop worrying about the ISO as it is just a tool for luminosity adjustment same as PP.

ISO is just a number for applied gain, and this can be done in PP, that's true. But my point is that this is not what's being done. The camera is set to default jpeg settings, and to whatever ISO is being tested, and then the exposure and/or illumination is adjusted until the histogram looks right.

And this leads to nothing of any use to anyone.

Performing all tests in RAW with identical light and exposure and then pushing the images in PP until histograms match, as you suggest, will be a much better comparative test. Unfortunately no-one does it that way. Right now only methods like dxomarks provide anything of value. DPR and IR test photos are worse than useless - they are actively misleading.

Jesper

 theswede's gear list:theswede's gear list
Fujifilm FinePix X100 Konica Minolta Maxxum 5D Sony SLT-A37 Sony 50mm F1.4 +2 more
Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads