Full Frame and Bigger Pixels vs. APS-C and Smaller Pixels - The Reach War

[Sporgon]

Hey, can you post some high DR wildlife images, please? I'd like to be able to visualise what you're talking about Like I said way back in this thread, I assume it's partly cos you're going for big mammals rather than tiny songbirds, but I may be wrong.

I'd be happy to. This image just got picked up for a textbook run:

One of the first things I look for, above all else in wildlife photography is "animal-scapes". I find them far more interesting than nostril or "trophy" shots. This often involves using shorter telephoto lenses. And it often involves shooting in less than ideal light, and in conditions like this (eight below zero). The bison were happy as could be, though.

This was one of the rare instances when my 7D did okay. The shot is ISO 200, and obviously I'm trying to get the sunset exposure nailed perfectly while bringing up the bison shadows later in post. You can't fix a blown sky, but you can lift shadows. Well, sort of. When lifting the shadows, the 7D annihilated detail with severe banding and noise. I was able to fix much of it with tedious post processing, but a superior sensor would have delivered a cleaner image capable of printing much larger. I had my 6D with me, but it was attached to a 300mm.

Don't get me wrong, I'm happy, but it could have been cleaner.

Come on Michael, stop pulling everyone's leg. You shot that on a D800 didn't you

 

[Lee Jay]

You seem to be viewing the discussion through a "brand lens". My comments about dynamic range are within the context of shooting wildlife regardless of brand. If someone can figure out how to get the dynamic range of the Sony sensors into higher ISO's, I'll be the first in line.

You'd need to reduce read noise to practically zero, and even then the problem is you might only be getting a couple of photons in each pixel in the darker areas of the frame which means you'd be really shot-noise limited.

Shot noise is effectively in the light itself. The only ways to improve it are to have better microlens fill factor (already nearing 100%), better QE (already over 50%), lower losses in the Bayer dyes (they've already done that so the next step would be to get rid of the Bayer dyes), or just getting more light (faster lenses, longer shutter periods).

 

[scyrene]

Hey, can you post some high DR wildlife images, please? I'd like to be able to visualise what you're talking about Like I said way back in this thread, I assume it's partly cos you're going for big mammals rather than tiny songbirds, but I may be wrong.

I'd be happy to. This image just got picked up for a textbook run:

One of the first things I look for, above all else in wildlife photography is "animal-scapes". I find them far more interesting than nostril or "trophy" shots. This often involves using shorter telephoto lenses. And it often involves shooting in less than ideal light, and in conditions like this (eight below zero). The bison were happy as could be, though.

This was one of the rare instances when my 7D did okay. The shot is ISO 200, and obviously I'm trying to get the sunset exposure nailed perfectly while bringing up the bison shadows later in post. You can't fix a blown sky, but you can lift shadows. Well, sort of. When lifting the shadows, the 7D annihilated detail with severe banding and noise. I was able to fix much of it with tedious post processing, but a superior sensor would have delivered a cleaner image capable of printing much larger. I had my 6D with me, but it was attached to a 300mm.

Don't get me wrong, I'm happy, but it could have been cleaner.

Thanks! It makes much more sense to me now. Landscapes can need a lot of DR, so an animal-in-context shot likely will, too. I usually prefer to have a much closer crop, but as I say, I don't do mammals - and I tend to go for 'textbook' composition, maybe because I'm still early in my career.

It's a moody shot. Very atmospheric.

 

[MichaelHodges]

Come on Michael, stop pulling everyone's leg. You shot that on a D800 didn't you

8)

 

[Marsu42]

Why? Because the moon covered the same absolute sensor area. There is a difference in pixel count between the two images, but overall, both sensors gathered exactly the same amount of light! That's the key there. There is no advantage to a larger sensor if you are not utilizing that increase in sensor area.

Thanks for the great post! Two additions here from my 60d/6d experience:

1. The more sensor coverage you have (with about the same mp ff vs crop), the more you can profit from future developments in noise reduction. DxO's prime shows the way, and I'm sure there's going to be more developments once even more computing power is available.

2. You're talking of reach for tele shots, with reach for macro there's the aspect of a) flight distance of animals and b) light occlusion by the lens (for available light or flash). That's why I still prefer my 60d for insect macros and the like even over my shiny new 6d. The 100L is as sharp as it gets on crop, so no advantage of ff here.

Are you shooting wildlife at ISO 100 and 200 on a regular basis?

Me, too, esp. because Magic Lantern's +3ev dynamic range boost with dual_iso only works with base iso ... that's why I'm regularly using 100/800 or 100/1600 is for wildlife for shadow/sun, sunrise/sunset or catching specular highlights.

 

[lexptr]

Thanks for the interesting test!
Sorry if I will repeat something that was already said (have no time to read all 10 pages...)
The results are not surprising for me. You have tested a very specific situation, where APS-C will show an expected advantage. An exceptional lens was used, capable to deliver a very fine detail, which was happily recorded by smaller pixels of APS-C camera. Hence you've got a better detail. It is actually very simple. And because shooting conditions was good (low ISO, moderate aperture) you had no problems with possible noise and diffraction. It is known that in some cases APS-C has advantages. Another example: some time it is better for macro, because you can get better actual magnification and deeper DOF.
But, anyway, thanks for another point for keeping APS-C camera, after I finally will upgrade to FF!

 

[jrista]

Thanks for the interesting test!
Sorry if I will repeat something that was already said (have no time to read all 10 pages...)
The results are not surprising for me. You have tested a very specific situation, where APS-C will show an expected advantage. An exceptional lens was used, capable to deliver a very fine detail, which was happily recorded by smaller pixels of APS-C camera. Hence you've got a better detail. It is actually very simple. And because shooting conditions was good (low ISO, moderate aperture) you had no problems with possible noise and diffraction. It is known that in some cases APS-C has advantages. Another example: some time it is better for macro, because you can get better actual magnification and deeper DOF.
But, anyway, thanks for another point for keeping APS-C camera, after I finally will upgrade to FF!

The quality of the lens in this case really doesn't have much to do with it. The resolution is seeing limited, which reduces resolution much more than even diffraction. On top of that, a 2x TC was used, which also reduces resolution below the diffraction limit. If the image was diffraction limited, then yes, the 7D is going to make better use of the detail being resolved by the lens. But that was not the case...I was seeing limited.

If I did a test where the lens was diffraction limited, the 7D would probably take a greater lead than I've demonstrated here.

 

[NancyP]

In case people don't understand "seeing limited", this refers to atmospheric interference when viewing celestial objects like the moon. Twinkle twinkle little star - the star is a point light source to us and should be a non-twinkling dot, but air turbulence, temperature differential, humidity in air all affect (refract) the image of that dot. That's why astrophotographers stack ("average") large numbers of identical moon or planet photos.

 

[Lee Jay]

In case people don't understand "seeing limited", this refers to atmospheric interference when viewing celestial objects like the moon. Twinkle twinkle little star - the star is a point light source to us and should be a non-twinkling dot, but air turbulence, temperature differential, humidity in air all affect (refract) the image of that dot. That's why astrophotographers stack ("average") large numbers of identical moon or planet photos.

Stacking is for reducing noise. The " lucky imaging technique" where you select the best frames from many (usually from video) is for getting around the seeing limit. You stack just those best frames to reduce noise on already high resolution images.

It's not an insignificant difference. In some cases, you can improve from a seeing limited image at 2 arc seconds to a diffraction limited one at better than half an arc second.

 

[mackguyver]

In case people don't understand "seeing limited", this refers to atmospheric interference when viewing celestial objects like the moon. Twinkle twinkle little star - the star is a point light source to us and should be a non-twinkling dot, but air turbulence, temperature differential, humidity in air all affect (refract) the image of that dot. That's why astrophotographers stack ("average") large numbers of identical moon or planet photos.

Stacking is for reducing noise. The " lucky imaging technique" where you select the best frames from many (usually from video) is for getting around the seeing limit. You stack just those best frames to reduce noise on already high resolution images.

It's not an insignificant difference. In some cases, you can improve from a seeing limited image at 2 arc seconds to a diffraction limited one at better than half an arc second.

Astrophotography seems very complicated - making the results that jrista and others share all the more impressive.

 

[jrista]

In case people don't understand "seeing limited", this refers to atmospheric interference when viewing celestial objects like the moon. Twinkle twinkle little star - the star is a point light source to us and should be a non-twinkling dot, but air turbulence, temperature differential, humidity in air all affect (refract) the image of that dot. That's why astrophotographers stack ("average") large numbers of identical moon or planet photos.

Stacking is for reducing noise. The " lucky imaging technique" where you select the best frames from many (usually from video) is for getting around the seeing limit. You stack just those best frames to reduce noise on already high resolution images.

It's not an insignificant difference. In some cases, you can improve from a seeing limited image at 2 arc seconds to a diffraction limited one at better than half an arc second.

Astrophotography seems very complicated - making the results that jrista and others share all the more impressive.

It's pretty complicated, for sure. It should be noted seeing limited spot size is rarely as good as 2" (arcseconds). Because of seeing in Colorado, it's usually somewhere between 3.8-4.1", even worse at times. In most of the US, seeing-limited spot size is usually a little over 3"...if you have a 2" seeing limited spot size, your seeing is pretty good. If you have seeing limited spots that are smaller than 1.5", your seeing is excellent. Assuming you use drizzling to increase the resolution beyond that, if you used full 3x drizzling, resolution could improve to 0.5". With seeing of 3.8", 3x drizzling might be able to improve resolution to 1.3" (you really need a LOT of frames to achieve that, though).

Also, getting diffraction-limited images at less than half an arcsecond is also rare. The longer the scope, the greater the magnification...very long scopes (when you add a 2x or 3x barlow, for example) are diffraction limited, but they magnify the spot so much that you can actually see the "waves", or the rings of the airy disk around the central star's peak (also only possible with truly excellent seeing...less than that, and the star jumps around and gets warped, so you still can't get a diffraction limited spot.) The central peak of a star might be 0.5", but the whole airy disk is still there an visible, so the actual diffraction-limited resolution is less than 0.5". With most telescopes, a diffraction limited star is larger than 0.5". With some of the best telescopes on earth, such as a Planewave or an RCOS, you might get on-axis diffraction limited spot at ~0.4", and off-axis (corner) diffraction limited spot at ~0.6". Those telescopes cost tens of thousands of dollars.

With a lens like mine, or your average astrograph refractors or RCs, your on-axis diffraction limited spot is usually going to be quite a bit larger than 0.5" in the absolute best of conditions. Any off-axis spots are going to suffer from some kind of aberration...astigmatism, coma, field curvature, etc. Corner spot size in many scopes can be quite large, and often looks like little comets or out of focus blur.

 

[Lee Jay]

Yes...the numbers I was giving were for the best of conditions. I've had seeing that was so bad it looked like I was looking through a jet engine wake!

 

[wickidwombat]

Why? Because the moon covered the same absolute sensor area. There is a difference in pixel count between the two images, but overall, both sensors gathered exactly the same amount of light! That's the key there. There is no advantage to a larger sensor if you are not utilizing that increase in sensor area.

Thanks for the great post! Two additions here from my 60d/6d experience:

1. The more sensor coverage you have (with about the same mp ff vs crop), the more you can profit from future developments in noise reduction. DxO's prime shows the way, and I'm sure there's going to be more developments once even more computing power is available.

2. You're talking of reach for tele shots, with reach for macro there's the aspect of a) flight distance of animals and b) light occlusion by the lens (for available light or flash). That's why I still prefer my 60d for insect macros and the like even over my shiny new 6d. The 100L is as sharp as it gets on crop, so no advantage of ff here.

Are you shooting wildlife at ISO 100 and 200 on a regular basis?

Me, too, esp. because Magic Lantern's +3ev dynamic range boost with dual_iso only works with base iso ... that's why I'm regularly using 100/800 or 100/1600 is for wildlife for shadow/sun, sunrise/sunset or catching specular highlights.

can you post a detailed description of how to use and process dual iso? ive tried it but the results are not good

 

[Marsu42]

can you post a detailed description of how to use and process dual iso? ive tried it but the results are not good

Well, you could have a look at the ML site but for a fellow CR regular here it goes

1. obviously: inststall ML
2. probably less obvious: enable "dual_iso" module in the ML menu

3. in the expo menu, enable "dual iso" and set the 2nd iso to use. I usually set the camera to fixed iso 100 and set ML to either 800 (=2.5ev gained) or when push comes to shove 1600 (=3.0ev gained). Then I expose for the highlights, i.e. I dial down ec to -2 or -3 until the highlights aren't clipped anymore - check the ML raw histogram for that.

4. you end up with an interlaced file DUALxyz.cr2 (if you've set the prefix option in the dual_iso menu). You then have to run a post-processing utility "cr2hdr" on the file to get a non-interlaced 16bit raw dng you can import into your postprocessing software.

Get cr2hdr.exe here (Mac/Linux also available somewhere): http://www.magiclantern.fm/forum/?topic=7139.0

The drawbacks of using dual_iso are:
a) it's harder to check for focus in camera because the image is interlaced
b) it's impossible to check for colors because they are screwed before processing with cr2hdr
c) postprocessing hassle, esp. time required for cr2hdr processing
d) dual file storage because you want to keep around the original cr2 in case cr2hdr receives further improvements
e) results usually needs manual wb (esp. tint) setting even though cr2hdr tries to autodetect it
f) if you use it regularly, you really wish you'd have a Nikon with native 14ev @base iso

All in all, I tend to use dual_iso very often, it simply saves me all the 2x brackets I used to do before and had to assemble with a hdr software, accepting the problem with frame differences like moving leaves or grass. I cannot believe how limiting it was to be stuck with the 10.5ev of my old 60d now that I can use 14+ev to capture shadows and bright sunlight in one frame.

 

[eninja]

Does anyone can list down "remarks and conclusion or rules in layman's term", targeted for people who got two bodies FF and APS-C to be able to get better quality photo?

To rephrase it, if one got two bodies but only one lens without the photographer moving closer to subject. which body should I use to conserve image quality?

If ISO to be use is less than 800. should I use 6D and crop it or I better use my 700D?
On the other hand, if ISO to be use is >1600. should I use 6D and crop it or I better use my 700D?

This thread is so picture quality geek into sub pixel level, i dont even know if my question is appropriate.

Thanks in advance.

 

[jrista]

Does anyone can list down "remarks and conclusion or rules in layman's term", targeted for people who got two bodies FF and APS-C to be able to get better quality photo?

To rephrase it, if one got two bodies but only one lens without the photographer moving closer to subject. which body should I use to conserve image quality?

If ISO to be use is less than 800. should I use 6D and crop it or I better use my 700D?
On the other hand, if ISO to be use is >1600. should I use 6D and crop it or I better use my 700D?

This thread is so picture quality geek into sub pixel level, i dont even know if my question is appropriate.

Thanks in advance.

To keep it as simple and generic as possible: When you are limited in how close you can get to your subject, and are using your longest lens, use the sensor with the smallest pixels.

That's all it really boils down to. It really doesn't matte how big the sensor is, if you can't fill the sensor, then sensor size doesn't matter. When you can't fill the sensor, the best way to extract the most IQ from your setup is to use the camera with the smallest pixels.

I don't think I can make it more layman's than that.

 

[HizerKite]

Sorry to reply to such an old post but just found this online and is very interesting and a great comparison. Last night I took 2 photos of the moon using a 400mm lens + 1.4 x extender. One on my 7D ii and the other on a full frame 1DS mk iii (Old I know). The 7D picture was considerably better - very surprised as images that I haven't cropped so heavily look a lot better on the 1D.

Really interesting to see your direct comparison