A7RV various RAW mode.

Started 2 months ago | Discussions thread
SrMi
SrMi Veteran Member • Posts: 5,778
Re: Compressed for events, Lossless for scenics

chrisfisheye wrote:

SrMi wrote:

chrisfisheye wrote:

SrMi wrote:

chrisfisheye wrote:

Jacques Cornell wrote:

chrisfisheye wrote:

Jacques Cornell wrote:

chrisfisheye wrote:

Jacques Cornell wrote:

charlyee wrote:

Hello all, I am new to the A7RV, exactly 3 weeks to the date as a matter of fact. I am still learning of course but i was wondering if some you can post, which RAW mode you use and why?

Thank you very much in advance.

For event work, I shoot Compressed RAW for the faster frame rate and smaller files. For scenics, I shoot Lossless Compressed for ultimate IQ. I thought I would shoot Lossless Compressed M (26MP) for event work, but decided not to because this file format is incompatible with Adobe's Denoise AI

Oh ok..

I have found information which confirms what you and SrMi say

A bit disappointing from Adobe, I am not the only one surprised that it works only with pure raw.

AFAIK, Lossless M and S can be processed with Adobe. It's just Adobe's Denoise AI that won't work. DxO PureRAW and PhotoLab won't process the files at all. This is because the files have been demosaiced in-camera and aren't actually raw but more like pseudo-raw. DxO also won't process Canon's M and S RAW files, for the same reason. You could always use Topaz Denoise to noise-reduce JPEGs processed from M and S RAW files, although I suspect that would be less effective.

On the plus side, reviews indicate that if you're shooting at low ISO and don't need post-production noise reduction, the M RAW files actually resolve more detail than native 24-26MP cameras.

and DxO's DeepPRIME noise reduction and I regularly shoot up to ISO 25,600.

Thank you.

Yes, I understand the reasons but this is still a bit unexpected . I would expect also jpeg to work.

I hope they will support other formats in the future.

I'm not holding my breath. RAW is raw, and quasi-RAW is not.

This is not the question.

Of course jpeg for instance can benefit from AI for noise reduction..

Yes, but the AI NR algorithms are completely different for raws than JPEGs. The powerful AI NRs of DxO and Adobe require non-demosaiced data.

Completely different, maybe you exaggerate a bit..

Check out this Adobe article for more details. The algorithms are trained to perform demosaicing and denoising in a single step.

Denoise demystified

When you see the examples, you understand that it could work perfectly with jpeg.

We understand that for the ultimate IQ, they decided to train it with raw, which is good of course but we can imagine they could have trained it from jpeg also.

I doubt this would be totally different things, especially when we talk about AI.. AI works like magic. Train it with raw, it works, train it with jpeg it will certainly work too.

Now, but this is just my opinion, I think it is a kind of elitism. I have the same opinion especially with dxo, I am not sure they want to put any effort for these "poor" jpegs.. You have to feed dxo only with raw !! they only want to show their algorithm when it reveals the ultimate power of their algorithm.

If they really wanted to support jpeg, I really doubt it would be difficicult, especially with AI. But do they want to support it ? That is another question....

I believe the reason to start with raw as input is that you get better results when you comb one demosaicing with NR as raw files have more relevant data.

Topaz Denoise allows both raw and JPEG/TIFF input and you get better results with raw input.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow