C200: Is Cinema RAW Light Going To Be an Orphan Format?

I'm thinking C300 Mark III too...and maybe something like a pro XC10/15.

Maybe even to the DSLRs/mirrorless' (although doubt it) as rumors have it Canon is going to step up their specs game:

"The source claims that Canon no longer wants to lose the “spec war” with their future camera releases. This way of thinking is trickling down to various Canon subsidiaries around the globe to prepare for the changes coming."

http://www.canonrumors.com/video-ca...-all-future-canon-prosumer-camera-bodies-cr2/
 
"The source claims that Canon no longer wants to lose the “spec war” with their future camera releases. This way of thinking is trickling down to various Canon subsidiaries around the globe to prepare for the changes coming."

http://www.canonrumors.com/video-ca...-all-future-canon-prosumer-camera-bodies-cr2/
The irony being that the cameras that loose the spec war are often the better and more expensive cameras than the ones that win the spec war. Canon is more expensive and in my opinion better than Sony (which won the spec war), and Arri is more expensive and better than Red (which won the spec war).
 
It's an opinionated discussion; I'm impartial and like them all.

But I think the FS7/FS5 and the mirrorless' clearly hurt Canon for the first time in probably ever (it was a slow process over 3-4 years, IMO), and if Canon does step it up, it's a win-win for us as all hell will break loose.

If ARRI didn't make a 4K camera (which it seems like they were pretty much forced to), I think they would eventually feel the impact too.
 
I'm thinking C300 Mark III too...and maybe something like a pro XC10/15.

Maybe even to the DSLRs/mirrorless' (although doubt it) as rumors have it Canon is going to step up their specs game:

"The source claims that Canon no longer wants to lose the “spec war” with their future camera releases. This way of thinking is trickling down to various Canon subsidiaries around the globe to prepare for the changes coming."

http://www.canonrumors.com/video-ca...-all-future-canon-prosumer-camera-bodies-cr2/

We will have to see if this happens. It looks like wishful-rumors that are so common today.
 
It wouldn't be like canon to implement a codec in a singular camera, they always spread out they're R&D cost over several cameras. That being said, (and speaking from no experience with either codec) I'd rather have internal Pro Res Raw than CRL on the next Canon C---.

On a completely other tangent, the voting system on this site is hilarious. Puredrifting, a backbone of this site for contributions and usable information gets down voted twice for asking a question, and starting a conversation?!?! Too funny, and now back to the regularly scheduled programming...
 
On a completely other tangent, the voting system on this site is hilarious. Puredrifting, a backbone of this site for contributions and usable information gets down voted twice for asking a question, and starting a conversation?!?! Too funny, and now back to the regularly scheduled programming...
I never vote on posts, but I just added the third down-vote to Puredrifting's post just to subvert the system and overthrow the hierarchy.
 
The SLR market is imploding.
So is the under Arri camera market.
Look at what RED just did.
RED wouldn't be doing all that consolidating if something wasn't going on.
And RED trying to get into the phone market.
Worrying about CRL being an orphan might be a worry that's not focused on the bigger picture.
 
It's not clear what the patent situation is, however a wavelet approach for RAW is better than a DCT or lossless/quantized approach (JPEG2000 encoding high-bit-depth monochrome would make sense for a RAW codec (beyond Redcode RAW and Cineform/GoPro RAW)).

Some folks are doing fine with 18:1 Red RAW (wavelet) whereas all the other RAW formats are huge. As compression goes up for wavelets, the entire image gets softer, with no other artifacts. For DCT methods, block artifacts show up, which is likely very bad for Bayer pixels, and thus DCT methods for RAW would likely be very light compression (if it's used at all). Given the file size and performance advantages of Apple's new ProRes RAW, Canon's Cinema RAWLite would best be dropped and for Canon to switch over to using ProRes RAW: https://www.apple.com/final-cut-pro/docs/Apple_ProRes_RAW_White_Paper.pdf . Canon has a relationship with Apple for ProRes support (C700), so ideally the license fees wouldn't block moving over to the superior ProRes RAW codec. C300 II / C500 owners can try out ProRes RAW today (Atomos Shogun Inferno).
 
It's not clear what the patent situation is, however a wavelet approach for RAW is better than a DCT or lossless/quantized approach (JPEG2000 encoding high-bit-depth monochrome would make sense for a RAW codec (beyond Redcode RAW and Cineform/GoPro RAW)).

Some folks are doing fine with 18:1 Red RAW (wavelet) whereas all the other RAW formats are huge. As compression goes up for wavelets, the entire image gets softer, with no other artifacts. For DCT methods, block artifacts show up, which is likely very bad for Bayer pixels, and thus DCT methods for RAW would likely be very light compression (if it's used at all). Given the file size and performance advantages of Apple's new ProRes RAW, Canon's Cinema RAWLite would best be dropped and for Canon to switch over to using ProRes RAW: https://www.apple.com/final-cut-pro/docs/Apple_ProRes_RAW_White_Paper.pdf . Canon has a relationship with Apple for ProRes support (C700), so ideally the license fees wouldn't block moving over to the superior ProRes RAW codec. C300 II / C500 owners can try out ProRes RAW today (Atomos Shogun Inferno).

I presume you know that Apple ProRes is DCT based compression? Apple hasn't said which compression algorithm they use for ProRes RAW, but I think it's reasonable to assume they use same type of compression as before, otherwise marketing would have made it a point to tell. It's Apple, so have to make conclusions based on what they don't say. BTW. At the NAB when Apple first introduced ProRes, I got to talk with one Apple rep who claimed he wrote the codec, and he told me it was DCT. But then everybody thought it was H264 based and Apple decided to just let people believe that and stopped telling people that it's DCT based.

I don't think DCT is necessarily such a bad idea for a RAW codec, but I don't know as nobody has done it before (except maybe Apple now...?). One thing I do know is that in very flat LOG and RAW codecs, more data is more data. The advantages of some of the more advanced codecs are lost when we really tweak a signal! Which we do with very flat LOG & RAW.

I doubt Canon is going to drop RAWLight. It will be used on many other cameras, and I think they would put it on C300mk2 & C700 if they could.
 
I presume you know that Apple ProRes is DCT based compression? Apple hasn't said which compression algorithm they use for ProRes RAW, but I think it's reasonable to assume they use same type of compression as before, otherwise marketing would have made it a point to tell. It's Apple, so have to make conclusions based on what they don't say. BTW. At the NAB when Apple first introduced ProRes, I got to talk with one Apple rep who claimed he wrote the codec, and he told me it was DCT. But then everybody thought it was H264 based and Apple decided to just let people believe that and stopped telling people that it's DCT based.

I don't think DCT is necessarily such a bad idea for a RAW codec, but I don't know as nobody has done it before (except maybe Apple now...?). One thing I do know is that in very flat LOG and RAW codecs, more data is more data. The advantages of some of the more advanced codecs are lost when we really tweak a signal! Which we do with very flat LOG & RAW.

I doubt Canon is going to drop RAWLight. It will be used on many other cameras, and I think they would put it on C300mk2 & C700 if they could.

I'm familiar with DCT (macroblock based discrete cosine transform, have implemented it in software for frequency-based (spectral) image analysis), as well as wavelet (full image (no macroblocks) DWT (discrete wavelet transform)). Due to the nature of the deBayering process, DCT type artifacts (discontinuous) will be more challenging than wavelets (simply softer), which will effect the final deBayered image quality. At light compression ratios with variable macroblock sizes (as with H.264/265 I-frames, vs. the fixed 8x8 macroblocks of JPEG) along with macroblock edge filtering, I can see DCT could work OK. Intuitively, inverse DCT (and associated pre-decompression (arithmetic coding etc.) and the typically very expensive macroblock edge filtering is an overall expensive process vs. wavelet decompression (which can be done in stages for real-time use = fast): I would not be surprised if Apple used concepts from the newer JPEG2000 (wavelet) for ProRes RAW vs. ProRes (based on MJPEG (DCT)). It might be possible to figure out what Apple is doing by using an (e.g.) SDI splitter and recording both uncompressed RAW along with ProRes RAW (not HQ: use the more compressed version) then performing a pixel difference analysis after deBayering to check for macroblock artifacts.

Canon's RAW Light is not competitive in file size or decode performance; if they want to provide the best solution for their customers, it would be in their best interest to switch to ProRes RAW, or develop a superior solution. DCT and wavelets are now very old methods; newer methods (which may be secret) could provide higher quality and less complexity = smaller files and faster decoding. For example, machine learning based image compression is relatively new, however it will provide massive improvements in quality in the next few years (in this case more complex, though custom chips will make this a non-issue): https://arxiv.org/abs/1611.01704
We describe an image compression method, consisting of a nonlinear analysis transformation, a uniform quantizer, and a nonlinear synthesis transformation. The transforms are constructed in three successive stages of convolutional linear filters and nonlinear activation functions. Unlike most convolutional neural networks, the joint nonlinearity is chosen to implement a form of local gain control, inspired by those used to model biological neurons. Using a variant of stochastic gradient descent, we jointly optimize the entire model for rate-distortion performance over a database of training images, introducing a continuous proxy for the discontinuous loss function arising from the quantizer. Under certain conditions, the relaxed loss function may be interpreted as the log likelihood of a generative model, as implemented by a variational autoencoder. Unlike these models, however, the compression model must operate at any given point along the rate-distortion curve, as specified by a trade-off parameter. Across an independent set of test images, we find that the optimized method generally exhibits better rate-distortion performance than the standard JPEG and JPEG 2000 compression methods. More importantly, we observe a dramatic improvement in visual quality for all images at all bit rates, which is supported by objective quality estimates using MS-SSIM.
You can see wavelet (JPEG2000) is superior to DCT (JPEG), and the new AI method is superior to both:
AICompression2.jpg

Machine learning compression/decompression* will likely need AI hardware and sufficient memory for databases, however already in 2018 we have AI chips that are quite powerful and low cost + low energy (currently used in e.g. DJI drones: Intel Movidius Myriad). Some cellphones have AI chips as well (e.g. iPhone X).

The point with respect to new RAW codecs is why use old, dated and/or non-competitive tech if creating a new format? RAW Light is already obsolete, so why waste time, money, and energy on it?


* really deconstruction/reconstruction, a very different method than typical compression methods like DCT and wavelet.
 
Last edited:
The SLR market is imploding.
So is the under Arri camera market.
Look at what RED just did.
RED wouldn't be doing all that consolidating if something wasn't going on.
And RED trying to get into the phone market.
Worrying about CRL being an orphan might be a worry that's not focused on the bigger picture.

True. I do think all of the mid-range (C200, EVA 1, C300 MKII and FS7) and GH5 level cameras will be outdated antiques very soon. They will replaced by phones or phone-like small, modular and cheap cameras realtively soon. I don't see Alexas being replaced by these devices but they definitely will eat away at the sub $10k cameras.
 
It's not clear what the patent situation is, however a wavelet approach for RAW is better than a DCT or lossless/quantized approach (JPEG2000 encoding high-bit-depth monochrome would make sense for a RAW codec (beyond Redcode RAW and Cineform/GoPro RAW)).

Some folks are doing fine with 18:1 Red RAW (wavelet) whereas all the other RAW formats are huge. As compression goes up for wavelets, the entire image gets softer, with no other artifacts. For DCT methods, block artifacts show up, which is likely very bad for Bayer pixels, and thus DCT methods for RAW would likely be very light compression (if it's used at all). Given the file size and performance advantages of Apple's new ProRes RAW, Canon's Cinema RAWLite would best be dropped and for Canon to switch over to using ProRes RAW: https://www.apple.com/final-cut-pro/docs/Apple_ProRes_RAW_White_Paper.pdf . Canon has a relationship with Apple for ProRes support (C700), so ideally the license fees wouldn't block moving over to the superior ProRes RAW codec. C300 II / C500 owners can try out ProRes RAW today (Atomos Shogun Inferno).

I have to say as a C200 owner, I would rather have Prores RAW in my C200. CRL is fine, nothing wrong with it but it's never going to become a defacto industry standard format like Prores RAW looks to be moving toward.
 
...I think they would put it on C300mk2 & C700 if they could.

C300 MKII, I agree. C700 and C700 FF, no way, CRL would be completely against their intended market which is a crew-served, feature or high end episodic user. That's why the C700s will only ever shoot RAW to Codex, they were designed that way. Codex is synonymous with studio-backed feature and episodic production, nobody else uses it really.
 
I agree, and believe CRM will eventually be phased out. However, it’s a big leap to make the assumption that ProRes Raw will become a standard just because ProRes is widely used. Currently almost every native raw camera has its own proprietary raw format and one can only assume that the large camera makers will only want to maintain their current business model as long as they can. Blackmagic and Kinefinity will likely incorporate ProRes Raw early on to drive sales but Canon, Sony, Arri, Red, Panasonic will take their time to avoid the licensing fees, etc. I’m interested in seeing some first hand, native ProRes Raw footage but until it’s available on a system besides FCPX, and only a feature for Atomos and DJI it’s hard to assume anything will change. I’d give it 5-7 years.
 
BM is supposedly (YouTube interviews) not interested in ProRes RAW and would like to mainly focus on CinemaDNG (at least right now).

I never see RED adapting it...REDCODE is brilliant and there is no point unless they are forced to by the industry.
 
BM is supposedly (YouTube interviews) not interested in ProRes RAW and would like to mainly focus on CinemaDNG (at least right now).

And now we know why: Black Magic launched their own RAW codec.
Interesting analysis over on CML: BM RAW codec is similar to ProRes RAW, both are DCT based, etc, etc, etc
Not sure I am allowed to link to it here.

In the theme of this thread: With more fragmentation in the RAW codec market and feet dragging support of ProRes RAW from Black Magic, I think Canon RAW Light will do just fine!
 
Back
Top