Big pixels or Little pixles?
- Thread starter Roger Clark
- Start date
Help Support BirdPhotographers.net:
Roger Clark
Banned
To be clear, the Moon as a test target is a very tough subject. There is color in the Moon, and there is a complete range of contrast, from very high (the craters on the terminator, to very low (sunlit areas away from the terminator). Note the subtle color variations in the basalt flows in the maria (the dark areas on the Moon). Note too, how those color differences are lost as ISO increases.
But this is no different than my test. Sensor size is irrelevant and the only/main factor changing here is pixel size (and the corresponding small variations in the different generations of the sensors). So for all practical purposes, the sensors could be exactly the same size here. That is why I only specified pixel size in the legends.
Roger
moon is a tough subject for sure and you have excelled in photographing it and getting to the bottom of what is best, it is just different compared to birds
I agree since your test was FL limited it doesn't matter how big or small the senor was, the same image was projected on all three sensors and it shows the effect of pixel size only.
My comment was more in terms "general" comparison between small and large sensors because one of the merits of the large sensor in non FL-limited situations is that you can get closer to your subject and get great IQ-your test is great for the subject and condition described.
I think the best is a large sensor with a pixel pitch that falls in the sweet spot of current generation technology
Last edited:
Steve Canuel
Well-known member
- Joined
- Feb 7, 2008
- Messages
- 5,444
Dumb question alert. Is it possible to build a sensor with two different size pixels and if so, would there be any benefits (best of light collecting and detail capturing).
Roger Clark
Banned
1st, yes one could, with a decent algorithm downsample the 7D image to be closer to the 5DII. At the high end, this will work quite well. In the deep shadows it doesn't quite work becuase of the added read noise in the pixels, which would add up to more noise than that from a single larger pixel.
But all this is a drift from the intent of the thread. With the Canon 1DX and Nikon D4 we have a choice of very large pixels and with cameras like the 7D and D800, a choice of very small pixels. But for people in focal length limited situations (many bird photographers have this problem), should one buy a larger pixel camera or a smaller pixel camera.
I didn't give my opinion because I didn't want to bias results for the first day. But here goes. I find your statement that the 5DII produces the best image interesting because I see it differently.
Here is my assessment:
In all the images, the 5DII images fail to show the subtle color differences that the 7D and 1D4 show. The color in the 1D4 and 7D are very close (until noise hides it).
ISO 100: 7D noise is small and detail is well above other images. 7D=top, 2nd=1D4
ISO 800: 7D noise is showing, but the detail is still well above the other cameras. 7D=top, 2nd=1D4
ISO1600: 7D noise is becoming prominent, but image detail is still very good. 7D=top, 2nd=1D4, but the difference is narrowing.
ISO3200: 7D noise is becoming objectionable and color is getting lost, in particular in Mare Serenatatis (the large circular dark area in the upper center). top=1D4, 2nd 7D. I do believe a good downsampling algorithm could improve the 7D to close that of the 1D4.
ISO6400: Noise is too apparent in 7D, and 5DII (which is slightly older technology than the 7D or 1D4). Top=1D4, 2nd=5DII. In my numerous sensor evaluations, I consistently see the 1D series sensors have fewer hot/bad pixels and the images here show that too: the 7D and 5DII images have a lot of "spiky" noise not seen in the 1D4 image.
In all the images, if you boost the low level, you will see that all the 7D and 5DII images have a lot of fixed pattern noise, which decreases as ISO increases. The 1D4 has a little fixed pattern noise at low ISO which quickly decreases at intermediate ISOs.
The bottom line for me: given a focal length limited situation and desire for as much detail as I can get, a camera with small pixels, like the 7D is the way to go. Not shown in the test, but what Arash alluded to: given a non focal length limited situation where you can change position to get the subject to fill the sensor, a larger sensor (e.g. full frame) with the most pixels is the way I would go. But if money were not an object, a compromise pixel size and high quality sensor like in 1D series is the way I would go.
Personally, with a small improvement in pixel efficiency implied by the latest camera announcements, pixels a little smaller than the 1D4 with the same signal-to-noise ratios as the 1D4 pixels in a full frame sensor would be ideal, something like 5-micron pixels (thus 34.5 megapixels) would be ideal.
Roger
I agree 5um @ FF is the best in overall IQ for a variety of conditions... let say the maximum bandwidth possible with current generation ASIC is equal to 1DX (14X18=252Mpixel/sec). For action the threshold is about 8fps this would give about 32Mpixls, pretty close to the figure above. So the ideal camera would be ~32 Mpixel FF @ 8fps.
Just imagine a 32Mpixel 8fps FF body with flagship AF in a compact and rugged body. This would have been such a hit covering from landscape to bird photography....bummer they didn't make it!
for now we have to just shoot with what we have, maybe next time
the way it is right now they make you buy at least two bodies so they can profit more...
Last edited:
Roger Clark
Banned
That is a different test and certainly valid. But to do that, one decreases or expands detail, thus the results are also dependent on the algorithm.
As in my response to Arash, the sensor size is irrelevant in this example. All three cameras could well have been full frame sensors. It is purely a test of pixel size and the trade of detail versus noise. Different people will have different opinions on which is better for them. There is no one right answer, much like in the film versus digital wars.
Roger
Roger Clark
Banned
Exactly. And in a an interview of Bill Gates by Barbara Walters, Bill said never produce a perfect product otherwise you can't sell an upgrade (that is a paraphrase, not and exact quote--I would like to fine the real quote). So if Canon or Nikon or Sony ... produced the ideal camera, we would not need to upgrade every couple of years.
Roger
The drawback is if you keep making small incremental improvements and throw away innovation you might as well become like Microsoft itself... :bg3: or maybe even Kodak at some point :w3
Last edited:
Michael Gerald-Yamasaki
Lifetime Member
Off topic alert.
Roger, Thought you might find this interesting. Grabbed it from a quicktime movie (& 'shopped it a little). It's from a Visualization Conference paper by David Akers, et al. It uses a method they call image-based relighting. Essentially image stacking where each image has different lighting... here the different lighting is different phases of the moon, using the high contrast edge area from each image.
See http://graphics.stanford.edu/papers/ib-relighting/
Cheers,
-Michael-
- Yamo
- Sat, 03 March 2012 7:19 AM
Roger, Thought you might find this interesting. Grabbed it from a quicktime movie (& 'shopped it a little). It's from a Visualization Conference paper by David Akers, et al. It uses a method they call image-based relighting. Essentially image stacking where each image has different lighting... here the different lighting is different phases of the moon, using the high contrast edge area from each image.
See http://graphics.stanford.edu/papers/ib-relighting/
Cheers,
-Michael-
Dan Brown
Well-known member
Is this different than an HDR?
Roger Clark
Banned
Hi Dan
Yes, because each image must be taken at a different time. A terrestrial analogy would be to start taking images of a landscape at sunrise and continue to take images for several hours and then somehow combine them to maximize the interesting shadow detail. This has been done with the moon for decades (so with film and darkroom methods, which must have been really tough). For a subject like the moon, these days a 3D terrain model is used and shading is added by computing a solar position that is ideal for each location, then drop color on top of that.
Roger
Michael Gerald-Yamasaki
Lifetime Member
Yeah. The images are taken over the lunar cycle - sickle, waxing, full (not used), waning, sickle - and only the high contrast area (long shadows on the craters) at the transition from light to dark is used from each image for the composite. If I recall correctly 12 shots (phases) were used. Similar to HDR, but it is the movement of the light rather than exposure that differs from shot to shot.
This was research in 2003.
Cheers,
-Michael-
Roger Clark
Banned
Hi Steve,
It is not a dumb question at all. There is one sensor with large and small pixels. I forgot who (panasonic?). But they have a neutral density filter over the small pixel so it is less sensitive and use it to extend dynamic range. But I suppose one could use it to enhance fine detail, though a new algorithm would have to be designed. Another method which is used on the Mars rovers is multiple sampling. For example take 10 images and because of vibrations each image is slightly misregistered from the others, and use those micro-differences to detect finer detail. Seems to work but I've never tried it myself. In non-focal length limited situations, it is easier to just increase focal length and do a mosaic.
Roger
Steve Canuel
Well-known member
- Joined
- Feb 7, 2008
- Messages
- 5,444
Thanks for the reply Roger.
Roger Clark
Banned
Here is an example of averaging 2x2 pixels. At ISO 6400, the 7D image is looking pretty ratty, but a 2x2 average really cleans things up and produces a very sharp pretty clean image. The detail is close to the 5D mark II, perhaps better in some areas. If the 7D were full frame with 4.3 micron pixels, then the full image would be 46.7 megapixels, and 2x2 averaged would be 11.7 megapixels. (Think of the D800 in this regard.)
Roger
Roger
Attachments
Michael Gerald-Yamasaki
Lifetime Member
It's interesting that the color noise remains, is even more visible in the 2x2 averaged version to my eye. For an image with a more saturated palette it would seem that color noise would continue to pose a problem, particularly if one intends to lift shadows in post-processing. Think of the D800 vs the D4 in this regard.
Cheers,
-Michael-
Roger Clark
Banned
Michael, Color noise is a lower frequency than luminance noise, so 2x2 pixel averages would have lass effect. but even so, I see the color noise much reduced, and significantly lower than the noise from the 5DII. Perhaps what you think is color noise is actual variations on the lunar surface?
I'm attaching a stretched version of my previously posted image to show the low end. It shows that the noise, both luminance and color is less than that from the 5DII, is one would expect. A 2x2 average of 7D pixels gives an equivalent pixel size of 8.6 microns, so larger than the 5DII pixels, resulting in better signal-to-noise ratios, thus our perception of lower noise.
Roger
Attachments
Last edited:
Michael Gerald-Yamasaki
Lifetime Member
The appearance of the noise in the reduced version to my eye is slightly more saturated while reduced in size. I sampled (PS) a few areas and did see a minor (1%) increase in saturation, but... there may be some spacial contrast differences from the size difference that I'm picking up visually or perhaps imagining. :w3
Noise being what it is, it's difficult to quantify small differences. Mentally interpolating sizes adds to the challenge. To my eye the 5DII 6400 image has less color noise by virtue of being slightly brighter. I'd be happier with a reduction that is done in conjunction with demosaicing rather than after. I can't parse the impact of conversion in relation to sensor pixel size. It would seem that camera profiles in the conversion process are skewing some of the information making it pretty difficult to know what is being compared.
Cheers,
-Michael-
ericbowles
Well-known member
Good thread and thanks for posting.
Just as high ISO noise performance is improving with each generation of camera, Dynamic Range and Color are also improving. At low ISO levels the newer cameras similarly outperform earlier models in terms of dynamic range and tonal range.
As suggested above, even the move to smaller pixels is offset by improved performance and processing.
