Cinelog - True logspace conversion for DNG and CinemaDNG footage

[Andy600]

Hi guys,

Please excuse the shameless plug (I hope the mods are ok with me doing this  )

We just launched a new product called Cinelog which works with Adobe Camera Raw and most raw video shooting Canon cameras, plus the Digital Bolex and Blackmagic Cameras.

Full info is on our website www.cinelogdcp.com

Unfortunately this release isn't a freebie (and yes we know there is something similar available for free) but Cinelog is true log space conversion for Magic Lantern Raw and other DNG based video and comes with a comprehensive LUT pack and guide.

If you have any questions or want sample frame conversions to test gradability let me know using the contact form.

[iaremrsir]

So is it calibrated for each camera's dynamic range, or is it a generalized curve?

[chmee]

Andy600
i'm interested in your subwords about the inofficial converters. tell, whats wrong and i try to fix that.

[Andy600]

So is it calibrated for each camera, or is it a generalized curve?

Each camera profile is calibrated specifically to their corresponding standard DNG matricies (2x color and 2x forward - in cameras that use them). The curve is logarithmic and gives equal results across all cameras (i.e. no clipped highlights or shadows and equal steps through all f-stops). Think similar to Cinestyle but better and for raw video with the benefits of Adobe Camera Raw debayering etc.

[Andy600]

Andy600
i'm interested in your subwords about the inofficial converters. tell, whats wrong and i try to fix that.

Hi Chmee,

It's currently impossible (or is for me at least) for Exiftool to change metadata with raw2cdng and Raw Magic. It just crashes.

These profiles require the exif tags 'Model' and 'UniqueCameraModel' to be populated to function correctly.

mlv_rec embeds the metadata by default but .raw_rec doesn't and footage needs to be retagged (easy to do and described 'how to' in the manual for mac/pc/linux).

If raw2cdng embeds both tags the Cinelog profiles will work fine. I'll amend the documentation and website immediately if raw2cdng has both tags embedded but it would be good if the DNGs can be edited with Exiftool in future as more tags are likely to be added to ML raw. It didn't with my older cDNG footage btw.

[iaremrsir]

Each camera profile is calibrated specifically to their corresponding standard DNG matricies (2x color and 2x forward). The curve is logarithmic and gives equal results across all cameras (i.e. no clipped highlights or shadows and equal steps through all f-stops). Think similar to Cinestyle but better and for raw video with the benefits of Adobe Camera Raw debayering etc.

So it's a complete color space? So if I were to take ML Raw footage and Blackmagic footage, and convert them to Cinelog there'd be no difference? (unless of course the scene's dynamic range exceeds the Canon's limit)

[Andy600]

So it's a complete color space? So if I were to take ML Raw footage and Blackmagic footage, and convert them to Cinelog there'd be no difference? (unless of course the scene's dynamic range exceeds the Canon's limit)

The colorspace is true log with DNG standard color which, in our tests gives a fairly uniform set of colors that reproduce very closely across all cameras (if the color matrices are set correctly correct). Intercutting footage from a BM and a Canon camera will be easy. There are 2 profiles per camera. One is has the DNG standard matrices plus our log curve and the other 'Cinelog CT' has an additional, built-in LUT. The website details the thinking behind the additional CT profile.

If you have a couple of frames from 2 different cameras I'll convert them with Cinelog and you can see what you think

[chmee]

ok. thanks for the hint. btw. with 1.4 i put a edit field for modeltag and unique model id into raw2cdng, but i think, i know, what thing causes the problem

[Andy600]

ok. thanks for the hint. btw. with 1.4 i put a edit field for modeltag and unique model id into raw2cdng, but i think, i know, what thing causes the problem

Great. TBH I converted most of my earlier footage with raw2cdng and loved the convenience of it... and the results

Support for .mlv would be the best thing

[avasarin]

Hi Andy600,
you write on the web site:
"Magic Lantern Raw Video that has been converted to CinemaDNG using Rarevision Raw Magic or raw2cdng is only supported for the 5D Mark III due to irregular metadata added by the applications."
So what if I have 5dmk2's DNGs converted with Rawmagic? Should I use 5dmk3 conversion package? Or I can't use it at all?
Thanks...

[kgv5]

Andy600, what is the difference between this and VisionLog? Is your cinelog gives any other results? Do you have any plans for making this work in resolve?

[Andy600]

Hi Andy600,
you write on the web site:
"Magic Lantern Raw Video that has been converted to CinemaDNG using Rarevision Raw Magic or raw2cdng is only supported for the 5D Mark III due to irregular metadata added by the applications."
So what if I have 5dmk2's DNGs converted with Rawmagic? Should I use 5dmk3 conversion package? Or I can't use it at all?
Thanks...

Raw Magic CinemaDNG's cannot have their metadata edited/corrected with exiftool so we cannot guarantee compatibility. If you want to send me one of your 5D2 DNGs converted with Raw Magic I'll check if it a) works and b) works correctly.

Cinelog profile's are unique to each camera model and will not work with other cameras so you would need Cinelog for the 5D Mark II.

[Andy600]

Andy600, what is the difference between this and VisionLog? Is your cinelog gives any other results? Do you have any plans for making this work in resolve?

Cinelog is a mathematically calculated log curve i.e. it's very accurate (it's similar to Log-C from the Alexa). It doesn't crush shadows, does not add any contrast to the image and retains all of the DR of your raw image. Cinelog also has full implementation of all color matrices useable by the cameras. The additional Cinelog CT profile also has a built-in LUT that improves on the standard DNG color palette and gives great skin tones on Canon cameras. Plus there is a nice LUT pack included

The Visionlog profiles crush shadows and lift the black point to make the image appear flat plus the gradations between f-stops are not equal. It uses a custom curve and isn't logarithmic. You might think this doesn't matter as it's still raw data but it matters a lot if you want to make the highest quality intermediates for color grading or inter-cut with log footage from different cameras.

[SteveScout]

Andy, don´t excuse, thanks for sharing the news. Just bought it. Curious to see how it stands up "against" VisionLOG and how good the overall dynamic range is preserved .. as I´m normally always working with Alexa ProRes footage on my day job I´m eager to see what you can squeeze out of the DNGs!

thanks!

Steffen

[Andy600]

Andy, don´t excuse, thanks for sharing the news. Just bought it. Curious to see how it stands up "against" VisionLOG and how good the overall dynamic range is preserved .. as I´m normally always working with Alexa ProRes footage on my day job I´m eager to see what you can squeeze out of the DNGs!

thanks!

Steffen

Thanks Steffen

We aimed to get close to Log-C gradeability, so as someone who works with Alexa footage I'm very interested to hear your thoughts. We made test curves that extend DR compression to ridiculous levels (but obviously they got very noisy), so we went with the most suitable one for the 5D Mark III's 11-12 stops.

[SteveScout]

Will come with feedback soon! We´re always shooting with Alexa or Epic-X as A-camera (our work: www.alexandsteffen.com), but I have my MK3 always with me to shoot inserts if needed in ML raw.

Intercutting the camera´s output after colorcorrection is never really a problem, but your approach could make it faster and way more accurate.

Stumbled across the first small drawback .. in my DNG sequences the metadata always reads "5D MK II" which I never understood why, so I can´t apply your DCP without changing the camera model with the exiftool. Great manual you made!

Have to check out DNGs with the latest builds. Or switch to MLV. Aargh .. so far the MLV workflow with converting them to RAWs first is painful, especially for longer projects. Need something like BATCHElor (with instant DNG and cineform offlining) for MLVs urgently ..

If this works fine I´ll implement cinelog in my "Raw camera workflow article" which - so far - features VisionLOG:   http://hackermovies.com/hackermovies-magic-lantern-raw-workflow-guide

cheers,
Steffen

[Andy600]

Will come with feedback soon! We´re always shooting with Alexa or Epic-X as A-camera (our work: www.alexandsteffen.com), but I have my MK3 always with me to shoot inserts if needed in ML raw.

Intercutting the camera´s output after colorcorrection is never really a problem, but your approach could make it faster and way more accurate.

Stumbled across the first small drawback .. in my DNG sequences the metadata always reads "5D MK II" which I never understood why, so I can´t apply your DCP without changing the camera model with the exiftool. Great manual you made!

Have to check out DNGs with the latest builds. Or switch to MLV. Aargh .. so far the MLV workflow with converting them to RAWs first is painful, especially for longer projects. Need something like BATCHElor (with instant DNG and cineform offlining) for MLVs urgently ..

If this works fine I´ll implement cinelog in my "Raw camera workflow article" which - so far - features VisionLOG:   http://hackermovies.com/hackermovies-magic-lantern-raw-workflow-guide

cheers,
Steffen

Thanks!

If your Mark III DNG's have the 5D Mark II tags, the meta needs changing definitely! If you used for instance the Cinelog 5D2 profile on 5D3 footage, the color matrix will be wrong i.e. wrong colors. We locked the profiles to the UniqueCameraModel tag deliberately to avoid wrong matrix calibrations being used.

BATCHelor and other raw2dng based converters usually keep the default models and these defaults can be whatever the app maker decided (or not changed at all). The raw2dng DNG meta implementation is basic but .mlv embeds more (and the tags we need for Cinelog). Changing meta can be a pain but hopefully the manual makes the process a little easier. Thanks for the compliment on the manual. I tried to be thorough

re: batch conversion. We are looking into development of a proper app for ingest and transcoding but it won't come anytime soon. I suspect you will see .mlv support coming soon in the current apps which should make things a lot easier. I just hope the app devs make use of the tags properly.

[Luiz Roberto dos Santos]

Great news Andy!
Just a note: I was looking at the data of your product ... in fact, seems to be amazing (like maybe a comparison with VisionLog, perhaps a mathematical analysis). That's when I thought about buying and came across the price of 29 euros!    His work is indeed very good, and I know exactly the work it takes to develop something that requires so much math. This even comes with LUT's correction... But do not you think somewhat expensive?
It's the same problem of "Cinegrain": the product is amazing, but absolutely expensive.
Furthermore, you are working with a digital product ... How long do you expect this to be distributed pirated form?

Remember of the Eastman Kodak to sell the Brownie. Sometimes we need cheapen to popularize it... 

[Andy600]

Thanks for your comment Luiz. I understand that if you compare it to something that is free it will seem expensive 

We developed Cinelog for several cameras and not just the Canons that shoot Magic Lantern raw video, and we are adding more. There is a full guide and the Rec.709 LUTs included but also 18 'look' LUTs that were developed using professional monitoring. I believe Vision Color sell their Osiris LUT collection alone for $69 (which I think is good value considering the work they did).

We have already received endorsements from several professional DIT's and colorists who are now using Cinelog as part of their raw pipelines because they understand the logic behind it and are confident in Cinelog's accuracy.

re: piracy - Do you think it would be less likely to happen if it were $1 instead of 29 euros? I don't

[Luiz Roberto dos Santos]

Exactly, but I'm in no way in favor of VisionColor. I, as a photographer and videographer indie do not have much money, I think the OSIRIS also somewhat expensive.
In any way want to belittle your work. Incredible actually think and reading texts on your website, I noticed that something is really accurate and well done. But one of the big reasons why we're here, in ML forum, is due to a market that has become popular, a lot, in recent years and this because, rightly, materials and solutions (aka ML) are cheap or open source for fantastic quality.
I know that if I want something of extreme quality, I will have to pay dearly, this is our capitalist world, but maybe prices that correspond to the market that it was created it would be fairer.
If you sell to producers who use Alexa, at this price, as colourists cited by you, they certainly will not question, because the "class" in which they work is much higher than ours, agree? Companies working with Alexa are the largest in the world, and will not ask to have it. The same can not be said for us indie.
I'm not telling you with the intention to do a fall in price, or something. The intention is only to question the producer who your consumers and adjust their prices to the same. Anyway, I hope to obtain the Cinelog future.

_{ps: I believe that would happen even if it was free.} 

[Andy600]

I understand your point of view completely Luiz. Cinelog certainly benefits professional users who may be working with Alexa and 5D Mark III raw footage but, being an indie myself, I wanted it to also be accessible for amateurs which is why we actually priced it very reasonably. Take for instance when Sony released S-log for their high end cameras. It cost thousands of dollars and basically does something very similar to Cinelog (although S-Log is for in-camera use). R&D costs time and money. Not everyone has need for optimum quality but there are those who certainly do and their reputations and businesses are built on it.

[iaremrsir]

I'm not sure how difficult this would be, but I think you can make this work for Resolve using 3D LUTs. So if we set the DNGs to REC709 or DCP color space, and set linear gamma, apply a CAMX2Cinelog LUT. And you have the speed of Resolve with your logarithmic curve for compression.

[Andy600]

I'm not sure how difficult this would be, but I think you can make this work for Resolve using 3D LUTs. So if we set the DNGs to REC709 or DCP color space, and set linear gamma, apply a CAMX2Cinelog LUT. And you have the speed of Resolve with your logarithmic curve for compression.

Already tried with linear profile DNGs and Cinelog embedded in a LUT. The results are not as good or as accurate as you get with the ACR workflow. You can get fairly close with a LUT if the embedded DNG metadata is correct but Resolve only uses the first DNG color matrix and it's DNG implementation is quite basic. We had issues with color casts, skin tones and other things that just don't happen with ACR plus ACR still beats Resolve's debayering 99% of the time.

We have a couple of pros putting some examples together and the differences are quite noticeable. If you use a batch script for rendering DNGs with Cinelog through AE it's comparable in speed to rendering intermediates from raw files in Resolve. I will post everything on our website when the content is ready.

[Danne]

Do you include the batchscript in AE to your lut?

[mageye]

I think from looking on your site this is interesting stuff . I am considering getting (buying) the Cinelog but even after reading all the information on the site I still am not sure how much of an improvement it will bring me?

I am by no means an expert, but I (like many), strive for the best possible image. I have used the VisionLog profile and think that I am following the correct workflow. Basically (if I am correct?), once you convert with Cinelog, you then use AE native colour adjustments (hue, levels etc.). I think I am right with that?

So I am a little confused already about how 'best' to treat the image here. What I would really like to see is a video or video tutorial that explains the whole process (including within AE) from beginning to end. I think what would also be pretty important, is to see how it compares to alternative methods (such as the 'normal' method through ACR) including VisionLog (perhaps a direct comparison?).

I really need to see a comparison of an image being manipulated through this so I can see what's going on and where it excels (which I am almost certain it does!).

I am someone that has to be absolutely convinced before I commit to buying something and basically I am still not certain as to how much it will benefit me and if it will improve image quality. The image is the most important thing and that is why I would purchase.

Thanks in advance