Screen resolution vs noise perception?

So, in trying to find answers and remaining open minded about my k3s performance in terms of noise and thd never ending argument about hi res cams ....

Would working with an older tech lower res screen on my 4 / 5 year old but still very usable laptop cause an increased perception of noise ??

I read an article this morning about retina display screens and seen some of my work modern large tv screens recently .... wondering if i updated ... yet another cost of technology. ... would i see less perceived noise ??

Also image quality at 100 and 400 % between k3 and k5iis was not noticeable .... would a better screen change this? ?

Roland and crew .... your thoughts

robbo d said:

So, in trying to find answers and remaining open minded about my k3s performance in terms of noise and thd never ending argument about hi res cams ....

Click to expand...

It is not clear to me from past discussions:

1. Are you shooting raw?
2. What NR are you doing?
3. Are you viewing files at the same size, or at pixel level?

robbo d said:

Would working with an older tech lower res screen on my 4 / 5 year old but still very usable laptop cause an increased perception of noise ??

Click to expand...

I wouldn't think so.

robbo d said:

I read an article this morning about retina display screens and seen some of my work modern large tv screens recently .... wondering if i updated ... yet another cost of technology. ... would i see less perceived noise ??

Also image quality at 100 and 400 % between k3 and k5iis was not noticeable .... would a better screen change this? ?

Click to expand...

It could, but what lens are you using? Consumer grade zooms are not always capable of showing resolution differences.

--
Dan

Dan,

I almost always shoot RAW.
I shoot k3 and K5iis side by side at weddings, both with high grade lenses, 17-50 & 50-135.
I have both cams set to NR off.
I download both sets of files on the same night to the same laptop.
I view them on screen and often then check out sharpness and noise at 100 or even 400% to make sure of eye sharpness.
I set both cams at auto iso up to 6400 for low light.
K3 is always a lot more noise, even in low iso let alone high iso and shadows.
K5iis has better DR which is also important to me.
....so im shooting in demanding situations a lot, pushing the cams to their limits, side by side. As the sun goes down and the shutter speeds need keeping up

I dont need to crop a lot and k5iis files at higher iso remain cleaner ... noticeably. .. (trying to avoid emotive language here), they also clean up better without undue waxwork smoothing.
I do PP in LR5 where i do my NR.

I sometimes find the k3 will meter and WB a scene better though and in that way it is a more reliable cam.

So .... wondering if my older tech screen was not coping with the noise ?

Always in hope that I find the magic bullet to prove my rants wrong, walk off with my tail between my legs and find new love with k3 as I did with my DA12-24 .....

Completely tech-less here, but I would be surprised if the answer to your question is yes. It would seem to be the other way around. If more resolution equals more clarity, then increased clarity should be more revealing. The size and space of the pixel is a known physical quality that effects noise. So, the argument becomes whether increased MP cameras have developed newer technologies sufficient to overcome this physics problem. Maybe they have, and maybe not so much.

My suspicions are that when you start messing with physics, you have to compromise somewhere. I am not personally drawn to the images of the larger sensor K3 (on aggregate). They seem less punchy, less intense, and less defined to my eye than images from my k5IIs cameras (not even considering noise). It is entirely possible that my eye is not refined enough, or that I am not used to a slightly different look. I had also, in some ways liked aspects of images from my K20D better than my K5 cameras. If Pentax offered a camera with the K20D sensor in an updated body, without the AA filter, I would purchase it.

Since I don't have a k3, I have relied heavily on photographers such as yourself who have both the K5IIs and the K3 for "unbiased" analysis. There are many who disagree with your assessment, and tend to base their disagreements on the technical advances of the K3, and then talk about per pixel performance that can be observed if one down-samples images --- etc.

I shoot weddings too, so noise, and low light AF will always be crucial in my camera purchasing decisions. Shooting a bird in a bright daylight sky will never answer my questions of how a cameras images will look shot at ISO 1600 or higher in low light. (It has always amazed me when certain reviewers test high ISO in outside daylight.)

DXO says the K3 noise performance is the same, or slightly better than the K5. Yet, plenty of photographers have noticed a significant difference in the noise performance between the two cameras. Are testing conditions unequal? Are photographers impressions too biased? Is there a lot of smoke and mirrors in photography, or just plain, down-right biased perceptions? Does any of this truly matter to clients?

The questions that Zvonimir Tosic has raised in his thread about MP madness are questions that I believe you and I share. The hammer hasn't changed in, who knows how long, because it is a tool well designed for its purpose. Now, if only we could get it with attached zoom lights for better target viewing, a bigger, but lighter head for better nail contact, an LED monitor to show how many strokes one is averaging per nail, the amount of pressure per stoke, etc, wouldn't our carpentry be so much better?

Yes, its all very interesting and frustrating....

I am relying on what i see, my eyes are pretty good and the results visually are real. I have been proven wrong before and no doubt will be again.

I am looking for the magic bullet....
??

I am looking to see if im doing something wrong but with both cams set identically and looking on screen as would any normal person, plus im using them often to their limits as you do at weddings, i see a telling story everytime.

Yes, they tell me the k3 has less noise ..... apparently read noise per pixel has reduced by about 25% but pixels have increased by 50 %...... dxo tells us colour depth and Dynamic range have reduced and their overall score of the k3 sensor is significantly less than the k5iis.

So when shooting in lower light, at 3200 or 6400 plus into sun, shadows, and all sorts of situations, i think those extra pixels and less dynamic range tells a big story.

Peoples skin instead of aeroplane surfaces are badly affected by high iso noise reduction and i hate the wax work look.

If my cams were in a studio with low iso, then im sure i could get stunning images, but as you know weddings are high speed and tests the gear.

I certainly get the feeling that im being told im dreaming, its all a figment of my imagination and i must be doing something wrong, as you suggest the specs tell a different story to mine..... perhaps non wedding shooters just dont get it ??

I think a modern take on the 16 would be mega and whilst living with the kx for several years, it was good but k5iis just rocks as an all rounder. Hence my leaning towards a nice 24mp FF in the near future.

Just looking to find any reasonable answer......

a friend just picked up the new d7200 and also has the d7000 with all high end Nikon lens.

he says the same the 16meg images are better even showing more detail as well.

cheers don

Hey there Don,

Trust all is well with you. Are you doing dance work still ??

I must admit i am having trouble seeing any reasonable difference when viewed at 100 - 400 %.with recent wedding shots. Skin detail was same in good light.

I have shot a couple of birds in good daylight, where the cropping was extremely good with the k3.

FF next stop, either D750 or if they give a 24mp option somehow with the Pentax. .... if it ever turns up ??? Lol

robbo d said:

So, in trying to find answers and remaining open minded about my k3s performance in terms of noise and thd never ending argument about hi res cams ....

Would working with an older tech lower res screen on my 4 / 5 year old but still very usable laptop cause an increased perception of noise ??

I read an article this morning about retina display screens and seen some of my work modern large tv screens recently .... wondering if i updated ... yet another cost of technology. ... would i see less perceived noise ??

Also image quality at 100 and 400 % between k3 and k5iis was not noticeable .... would a better screen change this? ?

Roland and crew .... your thoughts

Click to expand...

I would think that noise would be more noticeable using a lower pixel density screen. But I guess that depends on your vision and the viewing distance.

robbo d said:

Hey there Don,

Trust all is well with you. Are you doing dance work still ??

I must admit i am having trouble seeing any reasonable difference when viewed at 100 - 400 %.with recent wedding shots. Skin detail was same in good light.

Click to expand...

400% = interpolate 4:1 i.e your making up 3 out of every 4 pixels how do you expect to see increased resolution your not looking at the image but a made up version.

@100% every noise element will be 1.5 times the size of equivalent K5 image @100% so will show as more perceived noise (around .5 stops more noisy as DXO etc clearly shows)

But if you printed the k3 image would be 1.5 time bigger for the same detail.

or conversely if you view k3 files at the same size as equivalent k5 files your k3 image will have around .5 stops less noise fro any given ISO

robbo d said:

I have shot a couple of birds in good daylight, where the cropping was extremely good with the k3.

Click to expand...

Two simple rules

Crop = increase noise by crop factor

Sharpen before NR = increase noise algorithmically

robbo d said:

FF next stop, either D750 or if they give a 24mp option somehow with the Pentax. .... if it ever turns up ??? Lol

Click to expand...

The k3 and k5 have same size sensors but the k3 has 1.5 more pixels linearly so any niose will be 1.5 times smaller from the same size image.

i.e.

you can crop 1.5X and achieve very similar SNR figure k3-k5

you can not crop at all and achieve .5 stops noise drop

you can print 1.5 time bigger with the k3 and get same noise as k5

Because k3 noise elements are 1.5 time smaller you will need to change you noise removal work-flow to achieve optimum results.

I don’t think FF will help you unless you figure out what viewing size you need and what cropping etc will do to your image SNR

I FF camera without longer focal length glass will require cropping to achieve the same magnification in the image which will quickly negate any SNR advantage.

A 24mp FF camera with the same lens as a k3 will require a 2X crop which would destroy even 1.5 stop advantage

or put another way if you use 400 f5.6 lens on k3 you will need an 800 <=f8 to achieve the same noise figure as k3 does.

Sensor size is only one very small entitlement in creating low light noise free images.

Other things that help to lower noise floor

1 correct focal length (no cropping )

2 exposure

3 more pixel density

4 low noise electronics

5 flash

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
http://www.flickr.com/photos/awaldram/
1x.com
http://1x.com/artist/awaldram/wall

awaldram said:

I FF camera without longer focal length glass will require cropping to achieve the same magnification in the image which will quickly negate any SNR advantage.

Sensor size is only one very small entitlement in creating low light noise free images.

Click to expand...

A bit confused here. If a larger sensor creates less SNR, then why wouldn't cropping still be less noisy?

awaldram said:

Other things that help to lower noise floor

1 correct focal length (no cropping )

2 exposure

3 more pixel density

Click to expand...

Meaning more pixels on a sensor is better SNR? Wouldn't fewer, but larger pixels be better?

awaldram said:

4 low noise electronics

5 flash

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
http://www.flickr.com/photos/awaldram/
1x.com
http://1x.com/artist/awaldram/wall

Click to expand...

robbo d said:

Hey there Don,

Trust all is well with you. Are you doing dance work still ??

I must admit i am having trouble seeing any reasonable difference when viewed at 100 - 400 %.with recent wedding shots. Skin detail was same in good light.

I have shot a couple of birds in good daylight, where the cropping was extremely good with the k3.

FF next stop, either D750 or if they give a 24mp option somehow with the Pentax. .... if it ever turns up ??? Lol

Click to expand...

Hi Rob,

got a dance shoot coming up in a few weeks, been putting my new system together, new backdrop arrives soon, new lighting ect, new camera em5. will be shooting from the hip this one, not even going to use the vf, radio remote shutter release as well as face detect AF, cant wait, a good friend only 3 days ago picked up the new d750 $2.2k from hardly normal, he has dx and fx lens. went and had a play for 2 hours the other day nice camera same size as his old d70. crop sensor is only 10 meg though for his dx lens. I felt in operation it was no faster than my em5 for just about everything. low light is no revelation as everyone raves on about for FF, crop sensors are so good even at iso 6400 my em5 is amazing at fine detail. going next week for another play.

cheers don

Thanks for the clarifications.

robbo d said:

I almost always shoot RAW.
I shoot k3 and K5iis side by side at weddings, both with high grade lenses, 17-50 & 50-135.
I have both cams set to NR off.
I download both sets of files on the same night to the same laptop.

Click to expand...

All good so far.

robbo d said:

I view them on screen and often then check out sharpness and noise at 100 or even 400% to make sure of eye sharpness.

Click to expand...

But viewing 16mp at 100% is not the same as viewing 24mp at 100%. You are viewing pixel level noise, not what you get when printing an image or viewing both at full screen resolution.

robbo d said:

I set both cams at auto iso up to 6400 for low light.

Click to expand...

That is my limit as well.

robbo d said:

K3 is always a lot more noise, even in low iso let alone high iso and shadows.

Click to expand...

Viewed at 100%, I would agree.

robbo d said:

K5iis has better DR which is also important to me.

Click to expand...

The K-5 IIs only has higher DR at base ISO, i.e. ISO 80. At ISO 100 and above they are equal according to DXOMark, when viewed or printed to the same size.

robbo d said:

....so im shooting in demanding situations a lot, pushing the cams to their limits, side by side. As the sun goes down and the shutter speeds need keeping up

Click to expand...

I know what you mean, I've shot a couple of weddings.

robbo d said:

I dont need to crop a lot and k5iis files at higher iso remain cleaner ... noticeably. .. (trying to avoid emotive language here), they also clean up better without undue waxwork smoothing.
I do PP in LR5 where i do my NR.

Click to expand...

But is that viewed at 100%, or when printed? That's the part I'm questioning.

robbo d said:

Always in hope that I find the magic bullet to prove my rants wrong, walk off with my tail between my legs and find new love with k3 as I did with my DA12-24 .....

Click to expand...

I'm curious to know as well. Would it be possible to shoot at high ISO with both using the same lens, and upload the raw files? Or maybe we could both download ISO 6400 raws from DPR and do the conversions.

I truly believe that you are not accounting for the K-3's extra resolution, which allows heavier NR, nor for the fact that K-5 IIs files already have NR applied by Pentax.

--
Dan

Dan,

Perhaps the in cam (pentax) NR is far better than what we can achieve through 3rd party means like LR ??

Hence if i try applying some NR with k3.... mighr see better results.

Ill give it a whirl this weekend .....

robbo d said:

Dan,

Perhaps the in cam (pentax) NR is far better than what we can achieve through 3rd party means like LR ??

Hence if i try applying some NR with k3.... mighr see better results.

Ill give it a whirl this weekend .....

Click to expand...

I have never used in-camera NR on the K-3, but according to most tests I've read, it is not good. Try this free shareware.

Thanks Dan,

Ill give it a whirl this weekend.

Regards Rob

SBS said:

awaldram said:

I FF camera without longer focal length glass will require cropping to achieve the same magnification in the image which will quickly negate any SNR advantage.

Sensor size is only one very small entitlement in creating low light noise free images.

Click to expand...

A bit confused here. If a larger sensor creates less SNR, then why wouldn't cropping still be less noisy?

Click to expand...

A larger sensor - less magnifictaion

cropping = increased magnifictaion

Member said:

Member said:

Other things that help to lower noise floor

1 correct focal length (no cropping )

2 exposure

3 more pixel density

Click to expand...

Meaning more pixels on a sensor is better SNR? Wouldn't fewer, but larger pixels be better?

Click to expand...

Only @ Pixel level at image level more pixels - smaller noise source = finer less noticeable noise

This can be easily seen in the k3 and k5 the k3 is about .5 stop more noisy per pixel for any given ISO yet delivers about .5 stops less noise at the same image size.

This is because object signal tends to stay about he same i.e a 10x10 red square under the k5 becomes a 15 x 15 red square

But the noise is random so tends to not scale the same

i.e if the k5 had 5% noise the k3 8%

which leave the k5 with 95 signal units and the k3 207

The above is not technically correct but is an attempt to make a moderately complex idea easy to digest.

Member said:

Member said:

4 low noise electronics

5 flash

Click to expand...

Click to expand...

robbo d said:

So, in trying to find answers and remaining open minded about my k3s performance in terms of noise and thd never ending argument about hi res cams ....

Would working with an older tech lower res screen on my 4 / 5 year old but still very usable laptop cause an increased perception of noise ??

I read an article this morning about retina display screens and seen some of my work modern large tv screens recently .... wondering if i updated ... yet another cost of technology. ... would i see less perceived noise ??

Also image quality at 100 and 400 % between k3 and k5iis was not noticeable .... would a better screen change this? ?

Roland and crew .... your thoughts

Click to expand...

If you want to make an apples-to-apples comparison of a 16 MP sensor to 24 MP sensor, you have to use different magnifications. If you view a 16 MP image at 100% then you must view the 24 MP image at 81 or 82%. That will give you equivalent images.

Another approach is to view the 24 MP image at 50% and the 16 MP image at 62%. Again, you'll have equivalent portions of the image on you screen.

awaldram said:

SBS said:

awaldram said:

I FF camera without longer focal length glass will require cropping to achieve the same magnification in the image which will quickly negate any SNR advantage.

Sensor size is only one very small entitlement in creating low light noise free images.

Click to expand...

A bit confused here. If a larger sensor creates less SNR, then why wouldn't cropping still be less noisy?

Click to expand...

A larger sensor - less magnifictaion

cropping = increased magnifictaion

awaldram said:

awaldram said:

Other things that help to lower noise floor

1 correct focal length (no cropping )

2 exposure

3 more pixel density

Click to expand...

Meaning more pixels on a sensor is better SNR? Wouldn't fewer, but larger pixels be better?

Click to expand...

Only @ Pixel level at image level more pixels - smaller noise source = finer less noticeable noise

This can be easily seen in the k3 and k5 the k3 is about .5 stop more noisy per pixel for any given ISO yet delivers about .5 stops less noise at the same image size.

This is because object signal tends to stay about he same i.e a 10x10 red square under the k5 becomes a 15 x 15 red square

But the noise is random so tends to not scale the same

i.e if the k5 had 5% noise the k3 8%

which leave the k5 with 95 signal units and the k3 207

The above is not technically correct but is an attempt to make a moderately complex idea easy to digest.

Click to expand...

Excellent explanations. Thank you.

audiobomber said:

Thanks for the clarifications.

audiobomber said:

I almost always shoot RAW.
I shoot k3 and K5iis side by side at weddings, both with high grade lenses, 17-50 & 50-135.
I have both cams set to NR off.
I download both sets of files on the same night to the same laptop.

Click to expand...

All good so far.

audiobomber said:

I view them on screen and often then check out sharpness and noise at 100 or even 400% to make sure of eye sharpness.

Click to expand...

But viewing 16mp at 100% is not the same as viewing 24mp at 100%. You are viewing pixel level noise, not what you get when printing an image or viewing both at full screen resolution.

audiobomber said:

I set both cams at auto iso up to 6400 for low light.

Click to expand...

That is my limit as well.

audiobomber said:

K3 is always a lot more noise, even in low iso let alone high iso and shadows.

Click to expand...

Viewed at 100%, I would agree.

Click to expand...

So this is because i'm magnifying the image on a screen and getting the effect of more pixels ?? Even if the K3 read noise levels are decreased, I'm seeing the effect of the 50% more pixels ............

audiobomber said:

audiobomber said:

K5iis has better DR which is also important to me.

Click to expand...

The K-5 IIs only has higher DR at base ISO, i.e. ISO 80. At ISO 100 and above they are equal according to DXOMark, when viewed or printed to the same size.

audiobomber said:

....so im shooting in demanding situations a lot, pushing the cams to their limits, side by side. As the sun goes down and the shutter speeds need keeping up

Click to expand...

I know what you mean, I've shot a couple of weddings.

audiobomber said:

I dont need to crop a lot and k5iis files at higher iso remain cleaner ... noticeably. .. (trying to avoid emotive language here), they also clean up better without undue waxwork smoothing.
I do PP in LR5 where i do my NR.

Click to expand...

But is that viewed at 100%, or when printed? That's the part I'm questioning.

Click to expand...

Yes, Viewed at 100% .... not so much printing going on with me unfortunately. I'd love to do more .... starting to print a few more of my own. So printing gives a different effect than observing pixels on a screen ?

audiobomber said:

audiobomber said:

Always in hope that I find the magic bullet to prove my rants wrong, walk off with my tail between my legs and find new love with k3 as I did with my DA12-24 .....

Click to expand...

I'm curious to know as well. Would it be possible to shoot at high ISO with both using the same lens, and upload the raw files? Or maybe we could both download ISO 6400 raws from DPR and do the conversions.

I truly believe that you are not accounting for the K-3's extra resolution, which allows heavier NR, nor for the fact that K-5 IIs files already have NR applied by Pentax.

--
Dan

Click to expand...

So, going back to my original question ...., would a very high res screen which could cope with the pixel density of these modern hi res cams tell a different story ??

Interestingly I went back and looked at a few shots from the last wedding ... one was a detailed shot of a small chalk board sign with the bride and grooms name. It was shot at 3200 and it was very sharp .... had noise viewed at 100% but in this case admittedly was uniform smaller noise but retained extremely sharp detail of the chalk writing.

The K3 has improved Noise Read per pixel and is in many ways a technological improvement on the older 16mp sensor. (Even though Dynamic Range is slightly less ??)

But we have introduced 50% more pixels to get more detail and resolution.

With less DR, I would assume that increase pixel density in shadows would show up noise a little more ??

When viewed on most computer and laptop screens, the effect of noise is exacerbated due to screen resolution not keeping up with sensor pixel density.

Lower density sensors give the impression of looking 'cleaner' on screens.

When viewed at 100 +% we are magnifying the issue, but

This actually gives a kind of false indication of true noise / detail levels and when printed or on a very high res screen which could keep up with the pixel density would show a true result ... something which many people ....'fall into the trap of' ......... including me.

Many people have commented on K3 images and a recent post about a "mates D7200" shows that many people purchase these cams not understanding the issue fully, thinking they are getting the best available ... which they are .... however when viewing ...as most people do ..on screens and checking detail at 100% ... turn their noses up in disappointment and go back to their older cams in disgust. This is an inherent danger to the uninitiated who just look at a screen most of the time and go UUUGGHHH !!!

So, in reality these cams are more than most peoples needs and really are the domain of the specialist user or more tech savvy.

Most peoples screens don't help doing justice to physical reality of printing and actual noise.

Physics of poking 50% more pixels into the same area has it's issues, such as being more careful when hand holding in low light.

If I was going to shoot an image handheld in my K5iis at 1/100th, a K3 technically needs to be at 1/150th to get the same sharpness if we are being mathematically correct and therefore shouldn't be too hard to achieve. My concern was that in some circumstances that I may need to slip to a higher iso with my k3 to achieve the required shutter speeds ....

Another issue I have is that very few clients will print or print very few of their images. So many nowadays will plug their usb into the TV or laptop and view on screen. As screen resolution improves perhaps perceived results will improve?

This issue is not clearly understood by many. I am a photographer who enjoys composing and shooting, playing with the results on screen to get a pleasing image and thus far have been unimpressed with certain aspects of the 24mp sensor, which is difficult for many of us less technical to understand.

I think I am getting a handle on all this and I thank the team for their ..... dying patience .... Not withstanding the fact that I am leaning towards putting a 24mp FF into my kit, not for daylight performance, which I believe that Pentax aps-c is easily capable of delivering for wedding purposes, but that flexibility and extra stop or two once the light gets demanding.

I had a bride come around a couple of nights ago and we did a slide show on the TV screen and with both K3 and K5iis shots, showed some of the ceremony at 100-400% of her face / eye quality ..... she was blown by the quality and how big they could be printed.

In short it is a ....perceived...... increase in noise especially when viewing at 100%.

Am I getting closer ??