C70 Raw vs XF AVC Intra DR

[Vino]

Impressive result for the C70 at the CineD lab.
But why is the DR greater in XF AVC intra-mode than in the new RAW? Can somebody explain?

https://www.cined.com/canon-eos-c70-...-and-latitude/

(personally I havent filmed in raw much (used C300m2 before) because I film long docs and it felt complicated with work flow, Was thinking about starting to use it now, but why bother if the DR is like above?)

[digitalinnovations]

Could it be the lack of noise reduction in the RAW codecs?

[Maxbrand]

It's not uncommon and like Digitalinnovations says it has to do with the processing of the camera that ends up reducing noise in the shadows and therefore gives a better lab result for dynamic range.

[morgan_moore]

could I suggest bad science is occurng

yes one can reduce noise in the shadows but one will also be reducing resolution - ability to note changes in tone in those dark areas.

so less noise will come with a blurring of the image that is not being measured in the test

it is most unlikeley that the more compressed codec delivers some advantage at no corresponding cost.

It is obviously up to the user to evaluate if the loss of detail is worth the gain in lowered noise and also if the time savings of having ones NR done in camera are worth it. They may be, or not.

[morgan_moore]

a side obervation is that ' johnies face at 60%' is not a proper form of exposure as it delivers nothing on the wfm above 80% an is not consistent/repeatable.

clearly they (c5d team) didnt go to college - a grey chart should be used because johnie might get a tan in the summer and introduce inconsistencies.

SUch a reference point is not driven by poverty.. they have a whacking grey card in the scene to use if they choose.

[Clermond]

I'm not a big fan of these under/over exposing series.

I shot inside a car side by side raw clog2 and I can tell you with the C300iii it is quite some difference.

[cyvideo]

I'm not a Canon shooter but have found on Sony to get the most out of RAW you really need 16-bit Raw. Which you can get out most of the higher range cameras. For most everyday shooting I really see no major advantages over shooting RAW or SLOG over a good Custom XAVC internal Scene File. RAW can offer a number of advantages if given the time and correct workflow to massage the best out of it. Alister did a pretty good plus and minus advantages encoded vs LOG here.

https://www.xdcam-user.com/2020/11/r...g-does-it-too/

For you Canon users you may find this recent set of tests an interesting set of observations comparing the results between shooting XFAVC vs Canon RAW on the Canon EOS C300 Mk iii. Do any of you Canon users have any observations and comments on his tests and findings? I found his Canon observations quite interesting and his findings very much mirrored my observations in the Sony Linear 12-bit RAW environment. The bulk of which disappear in the Sony 16-bit RAW environment.

Chris Young

[Vino]

could I suggest bad science is occurng

yes one can reduce noise in the shadows but one will also be reducing resolution - ability to note changes in tone in those dark areas.

so less noise will come with a blurring of the image that is not being measured in the test

it is most unlikeley that the more compressed codec delivers some advantage at no corresponding cost.

It is obviously up to the user to evaluate if the loss of detail is worth the gain in lowered noise and also if the time savings of having ones NR done in camera are worth it. They may be, or not.

Thanks for answer everyone. Yes I got some doubt regarding testing too - or that reality is more complicated than I thought…

I would like to understand which codec to use to get the best possible DR out of the camera - also AFTER post production.
That is the most interesting.
If NR increases DR it might be relevant to do another test after that has been applied…

Regarding noise reduction and DR: Gerald Undone expose it in a RC talk and actually make some tests AFTER NR on RAW

https://m.youtube.com/watch?fbclid=I...ature=youtu.be

regarding detail there is also this article (same as Chris posted above)

https://www.cined.com/is-xf-avc-bett...-common-myths/

PS on a sidenote: I had forgotten that you might loose detail with Intra codec compared to Long GOP…
maybe there is no BEST codec, but different Codecs for different purposes… but that could the confusing in Post :-)

[morgan_moore]

I'm not a Canon shooter but have found on Sony to get the most out of RAW you really need 16-bit Raw. Which you can get out most of the higher range cameras. For most everyday shooting I really see no major advantages over shooting RAW or SLOG over a good Custom XAVC

video

I see that these vids dont really seem to me to really somehow 'get' raw.

Coming from a bground of shooting raw stills I 'know' where one needs raw..
to burn skies
to stop down and hold an essential highlight and the 'dodge up' the rest of the image
to perform a big colour swing
to apply a client defined look
to apply a different client defined look

im not saying raw is better (or worse) but I think filming a few trees doesnt demonstrate much (apart fron lines per inch)

so often when in a project there might be a few 8bit shots and as soon as you touch the rgb curves the colours just go bizarre in a way that doesnt happen at 10-12-14 (or 16) bit

[Grug]

Just to jump in on the conversation (not talking about these cameras specifically, as I'm not familiar with the C70 personally), but Raw is Raw. It's meaningless data until it's debayered.

The "compressed" recordings you get in cameras, are debayered images comprised from that raw sensor data. The "raw" images you see on screen in post-production software, are also debayered images (in this case, comprised from raw data that's been recorded).

So they're really not so different. Unless you actively make decisions in HOW the raw data will be debayered, that stray significantly from how you would have captured "compressed" versions of the same footage in-camera.

This is why every raw vs 10-bit+ compressed comparison test you've ever seen - has seemed to suggest a middling difference... because the difference is middling. It's the same raw data, there's only so many ways you can debayer it - ergo they generally look the same.

Where having that raw data does become handy, is when it offers greater bit-depth than your onboard compressed recording options do. Or when you stuffed up your recording a bit, and need to skew your white-balance significantly. Shooting underwater (for example) is one situation where raw becomes really helpful, because you don't necessarily have the ability to change camera settings as readily/easily as you would shooting on dry land. So having more latitude to nail your white balance in post, becomes a big incentive for shooting raw.

Outside of that? It's not a big deal. Any test that shows a compressed image scoring objectively higher (like these DR tests) is simply a result of in-camera processing skewing things a little - in this case with a bit of noise reduction. With raw files of the same shots, you can 100% match WHATEVER the in-camera recording is doing with a little time and effort. So there is no advantage to compressed over raw, it's simply the processing being applied.