Canon 7D MK iii

I really wish you guys would post some references and use correct terminology, or define the terminology you use.

John Sheehy wrote in post #18780070
The nature of gain is irrelevant, because only post-gain noise is affected by the type of gain, even if that mattered (and it generally doesn't, except when you use 32x mathematical gain, like the D5 with its highly-banded post-gain noise).

Gain through an electrical amplifier introduces it's own noise in addition to the noise in the original signal. Why do you say the D5 just multiplies the signal? They do no gain during the digitization process while reading the sensor? They only get millivolts from each pixel

John Sheehy wrote in post #18780070
Which is exactly what is needed to rid the comparisons of the diluting effects of noise reduction and equal photon noise (which is almost exactly the same on all Canon FF and APS-C cameras of the last several years).

Define photon noise. Do you mean "shot noise"? Thermal noise?

John Sheehy wrote in post #18780070
Yes, the 5D4 + 1.4x puts 20% more pixels on subject. It also lowers light levels ofr AF, and you can't do OVF AF with a 2x on the 5D4, though, when that means f/11 AF, while the 7D2 would use a 1.4x and still work at center-point, and we're really talking about the 7D3 in this thread, which I have no doubt about out-AF-ing the 5D4 at the same f-number.

We only got into high ISO and AF with the 7D2 because Mike used it as a proxy for the 7D3.

My point is that the idea of FF superiority at high-ISOs is overblown, if one isn't accepting (and enjoying) shallower DOF.

I could do 6D vs 7D2, and one can easily compare 6D to other FF cameras with many controlled images on the internet.

I would refer to the DPR studio comparison tool, since they "accidentally" shot the 5D4 at ISO 32000, which they usually don't do. If you compare the 7D2 at ISO 12800 to the 5D4 at ISO 32000, which is perfect equivalence (like a 1.6x TC on the 5D4), and whether you use the daylight or incandescent mode, both cameras seem to have the same visible chromatic noise in the dark areas of the image, in "Comp" and "Print" modes. When you compare the 7D2 to the R at 2x and 4x the ISO of the 7D2, the 7D2 is cleaner than the what you would interpolate the R to be at 2.56x the ISO. The R seems to suffer less than any other Canon in incandescent mode, though, with less blue-channel noise than expected, suggesting possibly a more transmissive blue filter. That doesn't help the outdoors shooter, though, where red is often the lacking color in ambient shade.

Here is another not real life test, I mean who shoots in a basement at such high ISOs at f4? Weddings are so much more real life with worse conditions... Of course this is a very terrible torture test and never would I do this in the real world, but at lower ISOs, I suspect I would see the same differences.

For John, here is the 7D2 at ISO 16000 f4, then the only difference is the 5D4 at ISO 32000 with a 1.4x TC. The 5D4 with a 1.4x and its pixel density actually provides a tiny bit more "reach" and I have to resize it down just a bit to compare to the 7D2.

Here are the two with no post processing other than taking the crop of the 5D4 and resizing it to match the 7D2 frame. The 5D4 is better than the 7D2, although both are crap at this noise level at 100% pixel viewing. I would only raise the 5D4 up 1 stop to match the TC impacts. This is what we would do in real life, when we are reach limited. Ex: Take a 400mm on a crop body or a 1.4x on a FF like the 5D4 and raise the ISO to match, then compare results.

I am not sure why the 5D4 shows a 1/3 stop better exposure, perhaps the teleconverter isn't really reducing the light by a full stop, but rather 2/3. It is consistently this way with this particular TC. Perhaps it is not really a 1.4x, but rather a 1.3x?

I've been away for a while. When is the 7D3 coming out?

May 17th

TeamSpeed wrote in post #18780629
May 17th

Don't you mean April 1st?

I am not sure what people are doing to make posts that have no reference to what they are referring to. I thought that even if you didn't include quotes, the header would say what post number or what other person the post is a reply to.

TeamSpeed wrote in post #18780101
I would NEVER use the 5D4 at ISO 32000, and I am probably one of the more proficient noise handlers and one of the more forgiving shooters.

That's not at issue. You're one of the best people I've seen at making noise take a subliminal back seat at high ISOs, without losing important edge detail.

My point is, with Canon sensors (and their lack of dual conversion gain, which complicates Sony sensors), there are only two fixed, immutable camera-specific read noises of note for short exposures; pre-gain read noise, which effects mainly deep shadows of medium ISOs and even higher tonal ranges with very high ISOs, and post-gain read noise, which is mainly what limits base-ISO DR, but generally becomes lost in the other noises and insignificant by ISO 400 to 1600, depending on camera, and is only significant at very high ISOs when the post-gain noise has spatially-correlated noise like chromatic blotchiness or banding noise, combined with heavy mathematical gain, such as what plagues the Nikon D5 at its very highest ISOs.

If we are interested in the difference in pre-gain read noise between the 5D4 and the 7D2, it only makes sense to go very high with the ISO, to reduce the diluting effects of NR and photon noise (which is basically the same per unit of sensor area in all FSI sensors of the last several years, within a very small range). It also makes sense to have no NR, or as little as possible.

If I am running a TC on a 5D4 to make up for the pixel density of the 7D2, it would be no more than 25600 in emergency situations, and would really want to keep the ISO at 16000 for that type of shooting.

All I can say is that a 5D4 slightly cropped shot with a 1.4x is more usable to me than a native lens on the 7D2 at a lower ISO. Whatever the theory is, those are real world results when I am out and shooting. Doing any more analysis in the real world where I second guess why I am not getting something closer with the 7D2 takes the fun out of it and is a waste of time. I have both cameras, and I use the tool that works best.

Perhaps you use a pixel-centric approach to NR and viewing the results. In any event, you have confirmed my main point, that there is little difference between the two cameras for pre-gain read noise, per unit of sensor area. DxO claims 1/3 stop less for the 5D4 (and the same photon noise, as is mostly true with all modern Canons), but I fail to see that in the DPR studio comparison tool, which gives a more raw chromatic noise when "RAW" is chosen. I assume that this is because the character or quality of noise (spatial distribution; not quantity) is better with the 7D2, which is renowned for the very random spatial character of its noise, even if not class-leading in intensity. I brought up the 5D4 because the DPR tool allows perfect noise equivalence, with 32000 shot for the 5D4. The 7D2 would own the R in high ISO noise, if it had 32000 shots, and you can see it if you set the 7D2 to 12800, and interpolate mentally between 25600 and 51200 with the R (except the the R seems to have some voodoo going on with incandescent; not helpful in outdoors and outdoor shade).

Again, what you have demonstrated is that even if the 7D2 is behind the 5D4 in equivalence, it is by so little that most people would never notice, if they actually did do an equitable comparison. This is very different from what people normally think of as differences of a stop or two, which they seem to think is true in practice, regardless of whether the practice is shallower DOF with the larger sensor with the same FOV, the same DOF and FOV, or focal-length-limited. To those confused people, "the larger sensor is just that much better across the board".

John Sheehy wrote in post #18780815
I am not sure what people are doing to make posts that have no reference to what they are referring to. I thought that even if you didn't include quotes, the header would say what post number or what other person the post is a reply to.

Citations, if you please. Something to back up your position. I see a lot of things being flung about as if they were fact (and they very well might be). I also see a lot of potentially questionable terminology being flung about too.

John Sheehy wrote in post #18780815
My point is, with Canon sensors (and their lack of dual conversion gain, which complicates Sony sensors), there are only two fixed, immutable camera-specific read noises of note for short exposures; pre-gain read noise, which effects mainly deep shadows of medium ISOs and even higher tonal ranges with very high ISOs, and post-gain read noise, which is mainly what limits base-ISO DR, but generally becomes lost in the other noises and insignificant by ISO 400 to 1600, depending on camera, and is only significant at very high ISOs when the post-gain noise has spatially-correlated noise like chromatic blotchiness or banding noise, combined with heavy mathematical gain, such as what plagues the Nikon D5 at its very highest ISOs.

John Sheehy wrote in post #18780815
If we are interested in the difference in pre-gain read noise between the 5D4 and the 7D2, it only makes sense to go very high with the ISO, to reduce the diluting effects of NR and photon noise (which is basically the same per unit of sensor area in all FSI sensors of the last several years, within a very small range). It also makes sense to have no NR, or as little as possible.

Again, what is this "photon noise?
What is "pre-gain read noise"? The CCD sensors got noise as the charge was shuttled to the amplifier/ADC, but CMOS sensors have the amplifier at each light detection element, making for a different sort of noise since each amplifier is slightly different. What are you defining as "reading"?

John Sheehy wrote in post #18780815
Perhaps you use a pixel-centric approach to NR and viewing the results. In any event, you have confirmed my main point, that there is little difference between the two cameras for pre-gain read noise, per unit of sensor area. DxO claims 1/3 stop less for the 5D4 (and the same photon noise, as is mostly true with all modern Canons), but I fail to see that in the DPR studio comparison tool, which gives a more raw chromatic noise when "RAW" is chosen. I assume that this is because the character or quality of noise (spatial distribution; not quantity) is better with the 7D2, which is renowned for the very random spatial character of its noise, even if not class-leading in intensity. I brought up the 5D4 because the DPR tool allows perfect noise equivalence, with 32000 shot for the 5D4. The 7D2 would own the R in high ISO noise, if it had 32000 shots, and you can see it if you set the 7D2 to 12800, and interpolate mentally between 25600 and 51200 with the R (except the the R seems to have some voodoo going on with incandescent; not helpful in outdoors and outdoor shade).

Again, what you have demonstrated is that even if the 7D2 is behind the 5D4 in equivalence, it is by so little that most people would never notice, if they actually did do an equitable comparison. This is very different from what people normally think of as differences of a stop or two, which they seem to think is true in practice, regardless of whether the practice is shallower DOF with the larger sensor with the same FOV, the same DOF and FOV, or focal-length-limited. To those confused people, "the larger sensor is just that much better across the board".

umphotography wrote in post #18780117
Ahahahhahaha

I would expect nothing less from some of the experts at POTN

Same song different day

Shoot bodies side by side in identical situations. No cars in some persons parking lot

compare noise results in side by side identical situations and then come back and talk to me. More importantly, shoot in not so good light or mixed light and then tell me about noise results with your 7D sensors

Your comparison seem to suffer from the fact that you don't recognize distortions of scale, and take things at face value.

In funhouse mirrors, perhaps. When you lose sight of the context of scale, as it relates to practical needs, you come to the wrong conclusions, and you never see those wrong conclusions contradicted because your observations always include loss of scalar context.

Well, there's your mistake. You probably should have never taken the 7D2 out, except as a backup, or for an extra shooter, if you had those other cameras. Does it sound like I'm contradicting myself now? I'm not. I never said or implied that a 7D2 can match a FF camera in all limited-light situations. If you thought that I said something like that, perhaps you project your fanatic tendency on others, and ignore subtleties and conditionals in their statements, pegging them as fanatics of something other than or the opposite of what you are fanatical about.

I've run into those real fanatics; a couple of m43 users once were 100% confident that the m43 frame is better than FF, because the same bird with the same focal length was a larger percentage of the frame with m43, regardless of pixel count or pixel density. You seem to be a fanatic on the other side from them. I am not a fanatic at all. I recognize the strengths and weaknesses of various systems, without fear of exposing the weaknesses of what I happen to own, or what I paid more for in inflation-adjusted dollars..

The 5D3 is kind of pushing it; the 7D2 has much better noise character, and less noise intensity in equivalence, and focal-length-limited. The 1DxII is better, the 5D4 about the same, IMO, but the 5D3 gives inferior in results, unless you are using the 5D3 in a way that capitalizes on a larger sensor, with shallower DOF, which, in the case of a wedding may be true, but don't push those shadows, though, because the 7D2 would have done better there..

If it got dark I would use the 5D3 instead of a 7D2. Why Unacceptable ISO results for our wedding needs and requirements.

I know what the bodies do in real world situations because I used them side by side to make the comparisons for almost 4 yrs.

... and perhaps you were looking at them all the while ignoring scale and purpose. You don't strike me as a person who fully grasps the practical meaning of ISOs in context.

It is irrelevant that a FF camera has less noise than an APS-C camera at the same ISO, because you do not use the same ISO or the entire larger frame to do the same thing, photographically. You only do when you accept the shallower DOF of 1.33 stops with lens choice on the FF, for the same FOV.

I like to think of a triangle of practical usage as a basis of comparison with different sensor sizes; the vertices are easy to talk about, but some usages will wind up interpolated in the middle somewhere:

1. Same FOV, and shallowest possible image DOF - the larger sensors have the lenses available to do this, and win here, but it is not possible without shallower DOF.
   
2. Same FOV, and same DOF - no generic benefit of larger sensors, or their ability to gather more light because of the greater area; this equivalence means less exposure on the larger sensor, making total light the same.
   
3. Focal-length-limited - no generic benefit to a larger sensor.
   

TeamSpeed wrote in post #18780155
Have you considered the Sigma 150-600? It will give you a bit more reach as the 100-400 with a 1.4x, however you can still use a 1.4x on that yet again for more reach then again. I think a 6D2 with that combination would be very good.

You'd have to use a non-reporting TC for f/9 (then seen as f/6.3), which has two disadvantages; the AF drive won't know that it might benefit from slowing down, and your EXIF info will be off. When the system does not see a TC (or an extension tube sometimes, too), focus can become extra-difficult.

These f/6.3 3rd-party lenses do not cooperate with a TC like the Kenko DGX, at least the older DGX that I have. I went out shooting a few days ago with my 7D2, Sigma 100-400/6.3, accidentally with my DGX instead of DG, and the TC was unusable. Yesterday, I took the DG out with me, and the combo was usable in good light (center point and 4 helpers, only, of course). The moral is, never assume that everything will work together, unless the TC uses protocol and reports properly, and combined f-number is legal on the camera. Any trick or combination of tricks against the protocol may backfire. Test before purchasing.

Archibald wrote in post #18780622
I've been away for a while.

Be ready to take cover. There is a tech slug-fest going on!

Perfectly Frank wrote in post #18780896
Be ready to take cover. There is a tech slug-fest going on!

LOL, I don't know if it is "tech". Maybe "technobabble"

Perfectly Frank wrote in post #18780896
Be ready to take cover. There is a tech slug-fest going on!

thank you for taking me out of this rabbit hole. crap I was in deep.

https://www.canonrumor​s.com …gets-another-mention-cr2/

It was always assumed there would be an R APSC body, through different stories, interviews and rumors. This is much of the same it seems, just kicks that proverbial can down the road a bit more.

TeamSpeed wrote in post #18780629
May 17th

ozziepuppy wrote in post #18780636
Don't you mean April 1st?

...of which year? All of these projections USWAGs about when launch will occur...