7D MkII vs 80D

TeamSpeed wrote in post #18357105
Pippan, if you would like to share your two raw files from the 80D, I will do the same type of comparison for you. It takes some time, but seems worth it.

Thanks TeamSpeed, I’ve put them in a Dropbox folder but I think you may have to PM me an email address to share it.

EDIT: I think you can access it from this link, https://www.dropbox.co​m …ZTbszSs_MOQc1Ad​V33ga?dl=0

I reshot them from a tripod in case it makes them easier for you to align, and uploaded the whole set from ISO 100 to ISO 12,800. BTW the reason I went from ISO 200 before (instead of base ISO) is that I’d been reading that the 80D is nearly ISO invariant from ISO 200 upwards so I wanted to see what difference there was between raising ISO in camera and lifting in post. I think from a noise point of view there’s little difference but when the banding kicks in it gets a bit ugly.

I will try to get to those tonight, I have to fill Invisicord orders first. The 80d strength is in the 100-800 ISO range, after that the tables start to flip for ISO performance.

EDIT: Here you go, ISO 200 pushed 5 stops on the left, 6400 on the right... I used ISO 200 and 6400 like on the 7D2. There is quite a bit of physical real dust on the color checker, so hopefully that doesn't distract from the comparison. The 7D2 has a cleaner ISO 6400 than the 80D, and it looks like you are better just staying at ISO 200 on the 80D and then later push/pull sections of the image during post.

TeamSpeed wrote in post #18357242
Here you go, ISO 200 pushed 5 stops on the left, 6400 on the right... I used ISO 200 and 6400 like on the 7D2. There is quite a bit of physical real dust on the color checker, so hopefully that doesn't distract from the comparison. The 7D2 has a cleaner ISO 6400 than the 80D, and it looks like you are better just staying at ISO 200 on the 80D and then later push/pull sections of the image during post.

Thanks so much Teamspeed, and how embarrassment! Yes the colorchecker was filthy . I've now cleaned it . Your comparison shows quite a bit less noise in the pushed image and, more interestingly, no apparent banding. I wonder if it could go another stop. I wonder if DPP deals better with that. I'll start playing with it.

I also wonder if there's any advantage of pushing in post vs raising ISO in terms of the detriment to dynamic range and colour depth.

Pippan wrote in post #18357361
I also wonder if there's any advantage of pushing in post vs raising ISO in terms of the detriment to dynamic range and colour depth.

This is the key with a (somewhat) ISO invariant sensor.

When we got an 80D at launch I shot a whole series of test images for the Photons to Photos website. It was fairly clear that (unlike previous Canon sensors) there's relatively little penalty for pushing. I later shot an image at ISO 6400 and pushed some of the shadow areas (likely to approx 25,600 equivalent). They looked no worse than just shooting at 25,600. Similar results were apparent across the range. It humiliated my 5D3 for a low ISO shadow push.

You'd definitely be better off shooting at ISO 6400 than ISO 100 if you need it, but unlike previous Canon bodies you could get away with ISO 3200 or 1600, or maybe even 800; then push in post (to ISO 6400 equivalent) without too much penalty.

What I've found (with the 5D4) is that I can rely much more on auto ISO, as minor underexposure isn't the issue it was on the 5D3. With the latter I used to get my initial shot with auto ISO, and if I had time I'd manually set the ISO; usually to push the exposure as "hot" as I could get away with. There's little need now for the effort of gaining an extra 1/3 stop to 1 stop by manual tweaks.

As you'd alluded to above - shooting at a lower ISO reduces the risk of clipped highlights, so ISO 800 pushed 3 stops in post will be OK in the shadows and will preserve 3 stops more highlight detail, vs just shooting at ISO 6400 in camera.

Note that this is true even of older Canon bodies when you're talking about two high ISO values (e.g. use ISO 6400 and push instead of 25,600; there's usually no noise benefit of 25,600, only clipped highlights).

In summary :-

• Pre 1Dx2, 5D4, 80D bodies: push the ISO in camera (unless you're already up at ISO 3,200-6,400 [depending on body] then just push in post)
  
• 1Dx2, 5D4, 80D bodies: No real need to push the ISO in camera, unless you would need to push in post by 4 stops or more
  
• Sony sensor users: stop laughing at us, we know you've had this for years
  

BigAl007 wrote in post #18357067
If you didn't use ACR, and that seems to be the case if you used Photo Ninja instead, then PS was working on RGB data, either 16 or 8 bit, depending on Photo Ninja, I don't know about it, I have never used it. So you have lost that big advantage of working with the RAW camera data. At the best it seems that Photo Ninja can do three stop of "exposure" so you would have had to add at least two stops to the RGB, and at worse if you had not added any to the RAW all five stops. These tests are always done using the RAW data, since it is there that you are working with the same data that the camera would use when doing "Digital ISO".

That does really depend on whether the likes of ACR keep a raw file in its original Bayer format for the purposes of tonal adjustments. I understand that some (Capture One?) can do some really clever things - such as infer the correct colour for a clipped highlight, by looking at surrounding pixels for the other Bayer colour channels.

However, it wouldn't remotely surprise me if Lightroom brings a raw into a temporary RGB style format (with some wide colour space); in order to allow the same set of tonal functions to operate on images for any camera. Otherwise they'd have to tweak the code for every different type of raw format (and any new functionality would need to be tested on all the old cameras).

In any event, converting a raw to a 16 bit RGB file with a reasonably wide colour space (e.g. ProPhoto) will result in some changes (vs the original raw); but I'd suspect it would be minimal in terms of judging the quality of pushed shadows.

I don't know the code of ACR or Lightroom, but I'm a coder (we're lazy - we don't like to code the same thing multiple times ) and I've also done a fair bit of pushing and tweaking of images in different colour spaces (so I have observed the minor shifts in colour behaviour).

Wow, thanks for all that info Sploo. Photons to Photos was one of the sites that led me to think pushing in post might be viable for the 80D. I got all excited by those tests and went out one night underexposing things way too much, and then discovered the horizontal banding thing when pushed too far . But as you say, 3-4 stops seems quite OK.

Without wishing to hijack this thread too much more, do you know why the 80D sensor maxes at 15,873 bits at ISO 100 and 200, but only 14,335 bits at higher ISOs than that (according to Rawdigger)? Seems like ISO 200 is some sort of cut-off level.

PS. I wish Photons to Photos would publish charts for the 100D! (I have one of those too)

BigAl007 wrote in post #18357067
If you didn't use ACR, and that seems to be the case if you used Photo Ninja instead, then PS was working on RGB data, either 16 or 8 bit, depending on Photo Ninja, I don't know about it, I have never used it. So you have lost that big advantage of working with the RAW camera data. At the best it seems that Photo Ninja can do three stop of "exposure" so you would have had to add at least two stops to the RGB, and at worse if you had not added any to the RAW all five stops. These tests are always done using the RAW data, since it is there that you are working with the same data that the camera would use when doing "Digital ISO".

Alan

I've been wondering about this myself so today I emailed Jim at Photo Ninja to ask what form the image is in when Photoshop does its thing after I've hit PS Done in Photo Ninja (having opened the file in Photoshop). Is it still raw data, or a jpeg, .psd or .dng, or something else?

He always replies promptly (just one of the things I love about this converter, great service), and said the image is transmitted back to Photoshop in rectangular pieces in a low-level internal binary format. Photoshop assembles the pieces into its internal representation in memory, and it isn't tied to any particular file format. I'm not sure exactly what that means but it doesn't sound like an RGB file.

sploo wrote in post #18357410
I understand that some (Capture One?) can do some really clever things - such as infer the correct colour for a clipped highlight, by looking at surrounding pixels for the other Bayer colour channels.

This is something that Photo Ninja does too.

sploo wrote in post #18357409
This is the key with a (somewhat) ISO invariant sensor.

When we got an 80D at launch I shot a whole series of test images for the Photons to Photos website. It was fairly clear that (unlike previous Canon sensors) there's relatively little penalty for pushing. I later shot an image at ISO 6400 and pushed some of the shadow areas (likely to approx 25,600 equivalent). They looked no worse than just shooting at 25,600. Similar results were apparent across the range. It humiliated my 5D3 for a low ISO shadow push.

You'd definitely be better off shooting at ISO 6400 than ISO 100 if you need it, but unlike previous Canon bodies you could get away with ISO 3200 or 1600, or maybe even 800; then push in post (to ISO 6400 equivalent) without too much penalty.

What I've found (with the 5D4) is that I can rely much more on auto ISO, as minor underexposure isn't the issue it was on the 5D3. With the latter I used to get my initial shot with auto ISO, and if I had time I'd manually set the ISO; usually to push the exposure as "hot" as I could get away with. There's little need now for the effort of gaining an extra 1/3 stop to 1 stop by manual tweaks.

As you'd alluded to above - shooting at a lower ISO reduces the risk of clipped highlights, so ISO 800 pushed 3 stops in post will be OK in the shadows and will preserve 3 stops more highlight detail, vs just shooting at ISO 6400 in camera.

Note that this is true even of older Canon bodies when you're talking about two high ISO values (e.g. use ISO 6400 and push instead of 25,600; there's usually no noise benefit of 25,600, only clipped highlights).

In summary :-

• Pre 1Dx2, 5D4, 80D bodies: push the ISO in camera (unless you're already up at ISO 3,200-6,400 [depending on body] then just push in post)
  
• 1Dx2, 5D4, 80D bodies: No real need to push the ISO in camera, unless you would need to push in post by 4 stops or more
  
• Sony sensor users: stop laughing at us, we know you've had this for years
  

Great info !!

I can see the science behind all of this but I would imagine a run/gun shooter is more concerned of "capturing the moment" with faster shutter speed in mind preventing motion blur. I think many events photographers would push the iso "in camera" and deal with noise in post.

Static scenery photography with monopod/tripods concentrating on composition I can see the luxury of shooting very low shutter speeds and reducing the ISO figures for the cleanest possible file. This type of photography is worlds different to fast moving subjects.

Oddly I do find I typically have better IQ with my 5dmk3 in events photography compared to my 80D.

With my shooting style I'd probably benefit with a 7dmk2 sensor in an 80D feature rich body

Does the "photons to photos" have a info on the Fuji sensors? I'm still more impressed with the X-T2 IQ over the 80D when it comes to available light shooting. Issue I have with fuji is flash photography for running and gunning (events) is still not something even remotely on par with Canon's system for AF accuracy.

Yes some charts have fuji data.

http://photonstophotos​.net/Charts/PDR_Shadow​.htm

There is so much to learn in this thread. Thanks for all the great discussion.

Thank you everyone for all this work. You guys are light years smarter than me in this respect. I barely understand the functions on my camera. So this is what I think your results say. 7DII push the ISO in camera for best results , and 80D use lower ISO and fix in Lightroom for best results. The 80D would make learn different patterns of camera use then I currently use. Right now I use AP then adjust ISO to get desired shutter speed. 80D I'm low light that won't work!

lijoec wrote in post #18357465
Thank you everyone for all this work. You guys are light years smarter than me in this respect. I barely understand the functions on my camera. So this is what I think your results say. 7DII push the ISO in camera for best results , and 80D use lower ISO and fix in Lightroom for best results. The 80D would make learn different patterns of camera use then I currently use. Right now I use AP then adjust ISO to get desired shutter speed. 80D I'm low light that won't work!

My take-away in how I use the 5D4 for example, is that I no longer have to worry so much about ETTR and pulling down bright areas. The camera is now more forgiving. I don't want to get "lazy" or "more carefree" about my settings, I still want to get the ISO close to what I should be using for the conditions, but if my images are a bit dark, or I have shadows that seem to be something I would have binned before, now I can relax and deal with them later. Hopefully I am right.

I will have to do this same test with my 5D4 to see how closely related it is to the 80D. What I see in the 80D example above is that it is better to use lower ISOs and push up areas than it is to use ISO 6400. ISO 6400 seems to be less clean than the 200 pushed 5 stops, however shooting color charts isn't very indicative of real-life shooting, so again my personal limit would be to shoot no longer than 2-3 stops below zero'd exposure and then push those up later, something I would typically never do with the old sensor.

Been reading along; this is interesting stuff. I have a real-world, low-light event to shoot tonight. At the last event at this location I was using a 60D. Parameters were generally around 1/200, f/5.6, 6400. This time I can take my 80D/70-200 f/4L IS (or 200 II). That gets me a bit more shutter speed.

I'm curious how those of you who know what you are doing would shoot this. Assuming I can get f/4 & 3200 or f/2.8 & 1600, how should I set up my camera to test pushing the 80D? How about dialing in f/4, 1/250 (or 1/320), & ISO 800, and pushing whatever I need in post for proper exposure? Do I dial in -2 (or whatever) EC to keep my shutter speed up? I think that my be easier than trying to manually keep the needle -2 manually. I will take a few shots at higher ISO (3200/6400) for comparison. I can't imagine needed a shutter any faster than 1/400.

I would use manual mode and set up a conventionally good exposure at a set ISO, like 3200 or 1600. Then set ISO to auto, and set your EC to -2 or so. The camera will then only be able to set the ISO down. Then test the post processing push to that of the higher ISO. I wouldn't think 2 stops though would really show much difference?

The tests really are useful with a wide DR scene however. As you raise your ISO, you lose DR. So a good test would be to shoot a high DR scene at ISO 400 vs ISO 6400, with the same shutter and aperture that was needed to provide you a good 6400 exposure. The post process the ISO 400 to push up shadows, and compare.

For what I shoot, the advantages at low ISO shadow pulling is only really useful for high DR scenes. Anything else requires fast shutter speeds for me, and that usually puts me out of the ISO 100-800 range that the new sensor really adds the extra usefulness.

http://photonstophotos​.net …rk%20II,Canon%2​0EOS%2080D

Now if you would like a bit of fun, add the 50D into that chart???? Then add the Fuji XT2.... Interesting where different models fit...