5dIII vs. D800 - just one nagging question

[Busted Knuckles]

I am betting Daniel has the answer to this - I think he knows more than he lets on...

In all the comparisons I have seen, the d800 wins big on dynamic range in the video mode... YET, no one comments on this except Dave Dugdale did in his piece.

So DB - does the extra shadow detail show up in the stills? Any dynamic range comparisons out there that worth any salt?

I was thinking of titling this thread as "2nd impressions are there any?"

[Daniel Browning]

So DB - does the extra shadow detail show up in the stills?

Definitely. Much more so than in the video, where you are limited by the 8-bit file format.


Any dynamic range comparisons out there that worth any salt?

I think this one is good: http://www.fredmiranda.com/5DIII-D800/index_controlled-tests.html
(http://www.fredmiranda.com/5DIII-D800/index_controlled-tests.html)
http://www.fredmiranda.com/5DIII-D800/t/canon-mk3-2b.jpg
http://www.fredmiranda.com/5DIII-D800/t/nikon2b.jpg


See how much cleaner and more usable the shadows are on the D800? It makes a huge difference for what I shoot.

I've done some dynamic range comparisons for myself (but only compared to the older 5D2, not the new 5D3) but haven't put them up yet.

[Busted Knuckles]

Thanks. I found the DxO rates the d800 about 3 stops greater range than the 5d3. (they seem to have done a decent job - or at least do a lot of comparisons) That strikes me as a quite a bit. The shots you provide are truly significantly different.

Again, thanks - any other twists and turns in the land of Nikonian that are either pleasant or unpleasant surprises now that you have had some decent view finder time?

Mike

[HDNitehawk]

Daniel; a quote from the review you linked,

"As you can see from my samples, I mostly stuck with the Canon Mark III on this trip. Mainly because of the difficulty I faced with Nikon D800's poor LCD Live View performance in low light. For my photography needs, this was the Achilles' heel of an otherwise superb camera."

Do you see this as an issue?

Fred Miranda was not to positive on the Nikor lens he had either.

[Daniel Browning]

Do you see this as an issue?

Yes, there are several problems with the liveview. Canon doesn't stop the lens down until you take the actual shot (unless you hold down the DoF preview button) and only *simulates* the exposure. Whereas Nikon is always simulating the DoF, which can cause noise problems for liveview. The way I've been working around it for now is to set the aperture to f/1.4 while setting focus (nice, bright liveview) then stop down to the actual shooting aperture. Perhaps there is something in the menu for changing the behavior to mimmick Canon.

The other problem with Nikon liveview is slow frame rates and row skipping. The frame rate slows down noticeably when you zoom in, and it's softer and more artifacted than a 100% view should be. My theory is that they the sensor readout speed is too slow in windowing mode (i.e. even when reading out just a small portion of the sensor) and they had to start skipping rows just to get it up to even the slow speed it is at. (If they read all the lines, then maybe it would slow all the way down to 5 FPS.)

[HDNitehawk]

I do not know if it was the D800's manual, or another tutorial put out by Nikon, but they had a whole section at the front on how to shoot in live view. Nikon seemed to put allot of emphasis on it, I assumed it was because it was going to be superior to Canon's.

[freelanceshots]

I am finding these results to be a stretch. I don't see noise like this at all in my 5D2 images and they are shot in the more extreme circumstances as far as dark areas can get and noise issues. Better yet its under the dreaded tungsten lighting here which makes images go poopy quickly. Image below captured with tripod, shutter 0.8, f/11, ISO 200, RAW file converted to JPEG for web at 90%. Crops are from the areas in Red. If the circumstance in ducussion involve taking this image and bumping it up in brightness by 2 stop from its RAW file I'd say that I've never come close to needing to do that at any time during my whole professional career. Maybe I'm not understanding this correctly as I'm not one of the brainy folks here. I have total trust in a 5DII and better yet a 5DIII. Most of the big companies I know have 5D2's and have no future wishes to jump over to nikon because its a non issue.

http://farm8.staticflickr.com/7085/7311284602_d7d5cd82ca_o.jpg

http://farm8.staticflickr.com/7244/7311235426_3c78ea249c_o.jpg

http://farm8.staticflickr.com/7217/7311236228_f1eab09714_o.jpg

A finished image

http://farm8.staticflickr.com/7130/6939251864_1353709162_c.jpg

[Daniel Browning]

I don't see noise like this at all in my 5D2 images and they are shot in the more extreme circumstances as far as dark areas can get and noise issues.

Personally, I don't consider these to be "extreme circumstances." For example, print film can achieve over 7 stops of highlight headroom (over middle gray). If you tried to do the same thing with the 5D2, you'd run into those noise problem.s But I don't think that means film photographers are/were extreme.

In any case, I'm glad that the capability of the 5D2 is enough for your needs.

[freelanceshots]

Long shutter speeds 8 seconds at ISO 250 in a dark room with a bright window with directional light and a white bedspread not difficult/extreme for a dig camera sensor? Try taking a wedding photo of the bride & groom with their white and black clothes standing infront of and off to the side of the window and without the use of strobes or a flash. Big difference between the black shadows/dark colors and the super blown out window.

Images that I shot last night. I don't see any downfalls using the 5D2 sensor where the 5D3 would have been even better from what I've heard people describe. Top image had shutter speeds reaching the 5 minute mark at ISO 50. Interior image used 1/5, 1/13 and 0.5 second shutters at ISO 640. I say you can nick pick anything to death where the time wasted could be better used doing something else. If my camera's images exhibited the detail and quality of Fred's test image then I would have dumped my 5D2 three years ago. Its all about how you choose to use your tool. Fred does have nice landscape photos so I'm not going to discredit him. Each camera has its weaker points but this signal to noise ratio aspect in respect to pushing the range of stops is not really that apparent unless you are trying to be unrealistic. That's all I'm saying.

http://farm8.staticflickr.com/7239/7313010318_0d53e83742_c.jpg

http://farm8.staticflickr.com/7098/7312375526_72f7730d5d_b.jpgpull this up from flickr to view larger: http://www.flickr.com/photos/freelanceshots/7312375526/in/photostream/lightbox/

[Keith B]

I'm sure interpolating up didn't help the 5D3 image on FM test. I think there are lots of folks upset Canon didn't go MP crazy and are trying to exploit every aspect of the sensor. Why not downsize the D800 image? When that would probably be more REAL WORLD.

No one is talking about how phenomenal the 5D3 is in low light.

[Busted Knuckles]

I think everyone is talking about the low light/low noise of the 5d3 and in the videos modes I have seen and are not talking about the dynamic range of the d800 except in two cases. Frame pulls are fairly lousy from both from what I have seen - but that is a true bastardization of the purpose of the video - neither Canon nor Nikon are wanting to be judged on a 30 fps 2mp frame rate, etc, etc.

My concern on dynamic range isn't during long duration shutter speeds, actually the opposite - what is the dynamic range in a 125, 250,500, etc. Where the dynamic range shows up in video is at 60th or 125th of a second - that is my nagging question.

If I understand it correctly ( and the last time I got this deep into I lost most of my gray matter when my head exploded w/ all the science/electrical engineering stuff) is that much of the low light noise comes from a heat and signal leakage across pixels that exacerbated by the long shutter time, etc. At a 60th/125th this shouldn't be such an issue.

I have expressed my desire to buy a 5d3 at the end of the summer - I currently have a t3i and the focusing capabilities look a lot (perhaps a little less) like the 5d2 - I haven't found that I missed focused that many times... yet. So one question is whether I would appreciate the bump to the 5d3 over the 5d2 (extra L lens).

net net - both these are fantastic cameras, completely beyond what has available a few years back - rivaling/approaching low speed film (100 iso) at light levels several stops faster.

FYI - I think the building shot is really quite good

[freelanceshots]

Tried video once and ended up having much better luck with my nephews canon $300 camcorder. I so wish Canon would make strictly just a kick ass DSLR and leave off the video. Got to be something useful that can fill its spot to improve still image quality or processing power even more.

[Jon Ruyle]

If I understand it correctly ( and the last time I got this deep into I lost most of my gray matter when my head exploded w/ all the science/electrical engineering stuff) is that much of the low light noise comes from a heat and signal leakage across pixels that exacerbated by the long shutter time, etc. At a 60th/125th this shouldn't be such an issue.


I'm not sure I know what I'm talking about, so take this with a grain of salt. But my understanding is that thermal noise is a major factor in long exposures, but that what really limits canon dynamic range at low iso is read noise, which is independent of shutter speed. For still objects, of course, one can get all the dynamic range one wants by taking multiple exposures.


I have expressed my desire to buy a 5d3 at the end of the summer - I currently have a t3i and the focusing capabilities look a lot (perhaps a little less) like the 5d2 - I haven't found that I missed focused that many times... yet. So one question is whether I would appreciate the bump to the 5d3 over the 5d2 (extra L lens).


IMO the main difference between 5DII and 5DIII is the autofocus. If you're happy with the autofocus of the t3i, you might be okay with the 5DII (I've never used the t3i, so take that with some salt, too). But if you shoot moving objects a lot, shoot very narrow DOF and would like to focus away from the center and not recompose, or want to focus in very low light, the 5DIII will make a world of difference.

[jrw]

have been following this thread for a while. It would seem that the big question in your mind is whether the dynamic range of the 5D3 is greater than the 5D2. The DxO tests put them in the same range. The D800 higher yet. The testing method is to downsize the images to a standard size for comparison. The similar resolutions of the two 5D bodies and the similar results show that Canon made no real progress in that department at lower ISO settings. At higher ISO, the 5D3 does pull ahead. The Nikon, with 50% more pixels, gains a little bit of noise reducing averaging as there are more pixels to mash together to create each pixel in the downsized image. This is how the rating of around 14EV can happen for a 12 bit file which at each individual pixel can only have a maximum of 12 EV. Both the Canon and the Nikon use 12 bit systems. In the real world we don't get all of it to use in our photos due to noise floor levels, the analogue amplification which adds noise, and the conversion process to digital which isn't perfect either. From the comparisons between the 5D3 and the D800 it would appear that Nikon emerges as the winner for less noise at the pixel level in the digital files that we wind up with. In low light (high ISO) DxO did show the 5D3 having greater dynamic range than the D800 but the 5D3 plateaued at settings under around 800 while the Nikon continued to improve as the ISO setting was reduced.

After looking over a lot of charts, graphs, and comparison photos I'll still continue to do as I have always done with digital cameras. Take a test shot, look at the histogram to see how the camera did, and walk away smiling with a proper exposure the first time, the second time with a compensated exposure, or if the dynamic range exceeds my sensors abilities a set of bracketed exposures. When I post process it is not very often that I adjust exposure and when I do, it is seldom by more than a third of a stop. I rarely have visible noise in my shadows from the 5D bodies. I will point out that if I push the exposure it does get ugly, especially in LightRoom, but not to the same extent as in Fred Miranda's posted shots. Yes, I did some major pushing in every piece of software I have just to see what happens. I do suspect that upsizing didn't improve the situation, but it is an area of ugliness that Nikon seems to have managed to not create in the first place. It is quite safe to say that the D800 does deliver better shadow details when pushing exposure.

[Busted Knuckles]

jrw - thanks. Real world we squish down and expand up all the time. Knowing what the performance of the sensor actually is, would give one a better understanding of the workability of exposures.

I also visited the DPreviews website and looked at the performance curves for several cameras - there didn't seem to be a whole lot of differences until you compared "native" performance (default amplification) with all the other settings that effect the dynamic range curve - then things moves around a couple of full stops - changing the slope of the curve - which if that is supposed to be good - why isn't it the default? I lined up the D800, 5d3, Nex7, and my humble yet tenacious T3i for a look at the DR curves. I wasn't sure what I was looking at most the time - but the curves for most of the setting didn't seem to stray from one another much.

With HDR processing coming into reasonable existence, reasonably static shots can be HDR'd and we reach/surpass the DR of film & our eyes?

I friend of mine is awaiting his d800 and I might just rent a 5d3 and go photoing with him and then compare images - before I change systems or plunk down $$ for the FF + FF normal zoom of some sort.

[Jon Ruyle]

With HDR processing coming into reasonable existence, reasonably static shots can be HDR'd and we reach/surpass the DR of film & our eyes?

I believe that with HDR and multiple exposures, DR is virtually unlimited. Though I read somewhere that the human eye has 24 stops of DR (who knows what that really means). If true, you'd need a about a 4000-fold range in exposure to match that with a DSLR.

[jrw]

I think you will be on the right track with renting and comparing bodies in the field. Florida is a little far away otherwise I'd like to be there and see how things stack up which would save you the rental fee. Mind you, I'd be looking over the whole user experience from turning the camera on to the finished print and heavily biasing my opinions with the unique perspective of my own quirks, oddities, and preferences. I recognize and respect that some people will consider bodies based on dynamic range, resolution, or any other feature that is of great importance to them, but for myself the entire range of lenses and accessories that I use would need to be evaluated separately and together with the body as it is difficult, for me anyways, to look at only one part of a system in total isolation.