Canon 5D Mark IV -- Time to Discuss!

filmuser wrote in post #18112411
http://flip.it/40Otpk
Why did they not hold off releasing mk4 untill this new design was ready for use?

Because the 5D4 is a stills camera and this would be for their cinema lineup.

filmuser wrote in post #18112411
http://flip.it/40Otpk
Why did they not hold off releasing mk4 untill this new design was ready for use?

That's still years off.

mike_d wrote in post #18111776
Ability to set any desired minimum shutter speed in auto ISO mode (5D3 max floor is 1/250')

That's pretty pathetic in the 7D2, also. It is very frustrating watching manufacturers limit things that are simple matters of programming that even a novice programmer could implement or change, if they had the source code, just to stick it to us to make sure we upgrade later. 1/250 is completely useless to me; I shoot jumpy little birds that turn into blur-fests at 1/250. IIRC, the 7D2 also has full-stop granularity with some or all limits (haven't looked in a while).

I found that very useful with the 7D2; on previous cameras, I had to switch from M to Av mode to use compensation when a sunlit white bird was against a dark, shaded background. I don't shoot for the JPEG, but ISO/EC compensation is also necessary for JPEG-dependent shooters.

This implementation of 4k is more like 30 4096*2160 JPEGs per second, which you can either use like a flip-book or separate out the frames. My aging Casio EX-FH25 does something like that; it can make a flip-book on the review image of up to 30 JPEG frames as fast as 40 fps at 9MP, and you get to choose how many of those frames are before the full shutter press, and how many are after it. This can be extremely useful, even though its not RAW - you can wait for action to happen, and fully press the shutter button afterward, and get the whole sequence. To do it with your choice of optics and a low-noise sensor is something the Casio can't do.

filmuser wrote in post #18112411
http://flip.it/40Otpk
Why did they not hold off releasing mk4 untill this new design was ready for use?

They just announced today having developed it, which probably means the patents were only just approved. They're probably years from actually getting it fully tested and integrated into a camera. This is going to be a major camera redesign factor involving a great many subsystems. Might see it on the next 1D.

ISO capability has climbed higher than I really need - anyone else long for a 'real' ISO 50?

cwphoto wrote in post #18112457
ISO capability has climbed higher than I really need - anyone else long for a 'real' ISO 50?

You mean an ISO 50 where you don't lose dynamic range thats subject to clipping highlights worse than ISO 100?

Sure. I'll take an ISO 25 while were at it. Shooting wide open I can find it's usefulness.

don1163 wrote in post #18112348
Just out of interest, does anyone know if it's possible to change the video codec by way of a firmware update ?
Maybe Canon could change it in a future update or Magic Lantern could come up with something..

Possibly. Real time video compression really needs hardware acceleration, so it would be dependant on the chips having enough performance to do the relevant processing. It's not impossible that their MPEG hardware couldn't process enough pixels at a high enough rate in order to do 4K MPEG, and if that's the case then no amount of new firmware will make it possible.

Charlie wrote in post #18112362
With a fast action sequences and a naturally faster camera, you will hit buffer limits more often. Faster cameras change your behavior a bit.

Indeed. I find that taking bursts is quite limiting on the 5D3, and waiting for the buffer to clear is a pain.

TeamSpeed wrote in post #18112373
That was true when I posted that link earlier, but supposedly now a file was found that had about 13.6 on it. More info will come out once this camera hits the mainstream testers. Lots of folks were hoping to get closer than 1 stop away from Sony DR.

I was hoping that the combination of newer tech, sensor size, and MP would give us something higher than the 1Dx2 in DxOMark "print DR" terms. Still, 13.6 would be nearly 2 stops of low ISO DR improvement over the 5D3, which doesn't make it class leading, but will make DR issues much less likely to bite in real world scenarios.

I've been meaning to ask this question for a while. I've heard from someone that ISO 125 is better than ISO 100 for canon cameras, is this true?

sploo wrote in post #18112461
Indeed. I find that taking bursts is quite limiting on the 5D3, and waiting for the buffer to clear is a pain.

I've never hit the buffer on the 5D3 when using CF only but using raw to both cards I have hit the buffer due to the SD limitation.

gjl711 wrote in post #18112397
It is interesting that after 5 years and multiple camera cycles, Canon has yet to reach the capabilities of the EXMOR, but 1 stop is pretty trivial.

exmor went backwards a bit recently in favor of high iso performance. I'm sure they have to tech to do both, but cant give out all the candy just yet.

filmuser wrote in post #18112411
http://flip.it/40Otpk
Why did they not hold off releasing mk4 untill this new design was ready for use?

global shutter has been talked about for a while now, just hasnt materialized for whatever reason.

John Sheehy wrote in post #18112433
That's pretty pathetic in the 7D2, also. It is very frustrating watching manufacturers limit things that are simple matters of programming that even a novice programmer could implement or change, if they had the source code, just to stick it to us to make sure we upgrade later. 1/250 is completely useless to me; I shoot jumpy little birds that turn into blur-fests at 1/250. IIRC, the 7D2 also has full-stop granularity with some or all limits (haven't looked in a while).

I am a bit confused, the 7D2 can set a minimum shutter between 1 second and 1/8000th in auto ISO settings.

From Canon:

FLiPxJB wrote in post #18112464
I've been meaning to ask this question for a while. I've heard from someone that ISO 125 is better than ISO 100 for canon cameras, is this true?

probably the wrong place for this discussion, but the phenomenon (real or imagined) is known as "Canon's Native ISO"

IMO you gotta be a pixel peeper of the highest order to worry about it.

TeamSpeed wrote in post #18111885
In the last 4 years, we have seen the release of the 5D3, 6D, 1DX, 1DX2, 5DS/5DSR, and now the 5D4 on the full frame front. We have only really increased the ISO performance from these bodies perhaps 1 to perhaps 1.5 stops (which I think is really pushing it). The pace at improving the high ISO has slowed down pretty dramatically. However with the last 3 cameras, Canon finally addressed (at least partially) the noise and DR issues due to the electronic noise in the shadows. That is actually quite significant, but perhaps is only a concern for a subset of the shooters vs those that desire cleaner high ISO.

There is still plenty of room for improving noise character (noises other than shot noise, as shot noise character is immutable, and very nice anyway). For very small extremely-high ISO images, the visible differences in read noise character can become tremendous for two cameras with the same DxOMark score. The D5 is a good example; it has very low high-ISO noise DxOMark-wise, but the character is pretty poor. You see more large patterns and little randomness when you downsample ISO 3.28 million to a small web size. It would be much, much better if it had totally-random read noise.

I suspect that going forward, there will be less and less ISO gains, just tweaks to what we have today, and instead all efforts will be on the features of the different models. Perhaps they can work out a real 25 and 50 ISO level instead, and start going the other way. Bring on ISO 12....

Light-based ISO 12 (IOW, ISO 12 that isn't a 3-stop ND of ISO 100) would be possible if pixels could be reset very quickly after reading them, to create cumulative exposure, without visible blackout in exposures with motion. You could get one RAW file from the cumulative exposure, or all frames might be saved so that you could leave out just one, like one where someone else's flash went off. Most current implementations of added exposures ("night" modes, IS extenders) have significant blackout, though.

Never say never to further decreases in read noise (approaching or reaching photon counting), or improving of QE. These QE figures that you see bandied about for current sensors are not the true full-spectrum sensor QEs. They are the QEs of green light, in the green-filtered pixels. The real QE of the red channel is typically only about 6% to 7% in the most efficient Bayer sensors! These things, if they happen, will happen pretty much in a big way, all at once, when they happen. Until then, we will mostly enjoy mainly improvements in character with incremental noise quantity a la "standard deviation".

TeamSpeed wrote in post #18112373
That was true when I posted that link earlier, but supposedly now a file was found that had about 13.6 on it. More info will come out once this camera hits the mainstream testers. Lots of folks were hoping to get closer than 1 stop away from Sony DR.

13.5 is around where most Sony bodies are. Some are higher, some are lower, but on average....

thenextguy wrote in post #18112481
13.5 is around where most Sony bodies are. Some are higher, some are lower, but on average....

That's a fair point - DxOMark have the a7Rii at 13.9. The D810 is still way in front though: https://www.dxomark.co​m …vement-in-DR-at-high-ISOs