Thank you for very solid testing. This confirmed what I already suspected. Having done work as a colorist for a while, it really does not cross my mind if footage is raw, 10, 12 or 16 bit anymore. It all works the same to me. I don't do much chroma keying, so I might have had a difference experience if I was doing more VFX.
@LaurentFranchet
2 ай бұрын
The bit depth makes a huge difference (16 bits RAW is linear and equivalent to 12bits log RAW and indeed, as clearly visible in this video, the XAVC codec is horrible for greenscreen work. In addition the FX6 can output 16bits RAW in 4264 x 2408 via HDMI. This higher resolution is a great way to improve the quality of the chromakey.
@jessecumberledge4530
9 ай бұрын
I have always said BRAW is Blackmagic's greatest feature. It's just the best codec by far. Edits as easy as prores, but with RAW capabilities and has smaller files sizes if needed. Big reason why I would choose a black magic camera vs others.
@elvisripley
10 ай бұрын
One of the things I have noticed with the Sonys is the XAVC I (Class 300) datarate is too low. Recording ProRes 422 HQ is enough for almost everything. But also XAVC I Class 480 is also enough. I have done tests recording raw and ProRes 422 HQ and then converting both of those to XAVC I 300 and 480 to check. If Sony added the higher data rate all software would already support it and be a nice option for green screen applications. I have little hope it will be added. Maybe in the FX6 and FX9 v2 we will get an H265 codec that has enough data to keep the compression artifacts to a minimum. The Burano has a higher quality XAVC H that might come to other cameras.
@chadterpstra
5 ай бұрын
I second this change!
@drew_hewitt
10 ай бұрын
Wow, I didn't realize the gap was so close. Makes me feel about better about shooting internally all these years.
@LaurentFranchet
2 ай бұрын
this video is VERY misleading. A RAW recording will always be much higher quality than an XAVC 4:2:2 10bits internal recording, no matter how you twist facts and images in a voutube video to pretend otherwise.
@deadendboredom
2 ай бұрын
@@LaurentFranchet I'm not sure if you're aware, but I am actually very critical of the all-I codec. It frustrates me that many people claim it’s the best recording option in the camera. How one evaluates internal recording versus RAW external really depends on individual needs and expectations. I've tried to objectively highlight both the pros and cons for different users-whether they don’t have extremely high demands or are those who scrutinize every pixel. Could you explain exactly which statements you find misleading?
@OlegUstimenko
3 күн бұрын
Preconceived notions: - there will be no tangible benefits when using arriraw over prores444. - cndg will have no tangible benefits over braw. might look sharper but that might be questionable interpolation by the nle. - a 10 bit 4:2:2 codec will be slightly worse at everything at the extremities compared to any 12bit codec, but will behave similarly in most other contexts.
@shueibdahir
8 ай бұрын
Also i have a Rebel T5i/700D and it shoots uncompressed 14bit RAW video at 2.7-3.8MP (your ArriRAW is 4.6MP) and the way the CDNG files behave are exatcly identical to ArriRAW. I remember being in complete awe when i graded the arri files in the past and its flexibility. The only raw i've ever seen come close to that is magic lantern raw. It's 11 stops of dr but can be boosted up to 13.4 usable stops at SNR of 2 when using dualiso. The problem with braw as i've used it is that its debayered in camera. Resolve doesn't fully do it. It's excellent for good quick results but only if you get the look you want on location. It doesn't have the same manipulation capability as uncompressed raw. The white balance manipulation is especially lacking in it.
@deadendboredom
8 ай бұрын
Have you conducted specific tests that demonstrate the difference or the advantage of ARRI RAW / uncompressed RAW from Canon and the disadvantage of BRAW? I would be very interested in seeing such tests.
@deadendboredom
8 ай бұрын
I just noticed that you seem to be from Germany. If you happen to be near Hamburg, I'd be more than happy to meet up and conduct tests with various cameras in situations where, in your opinion, the differences, advantages, and disadvantages can be best documented.
@nozer88
10 ай бұрын
Very interesting deep-dive. Thanks for the work!
@chadterpstra
5 ай бұрын
I’ve found in my tests of the FX3/6 cameras much more chroma blocking and loss of detail in the All-I recordings vs RAW. FWIW
@NMEdiary
9 ай бұрын
I think the internal recording format will have some baked in Noise Reduction. And that the Image will still be really good.
@deadendboredom
9 ай бұрын
I'm with you, the All-i image looks very good, and I hope I've communicated that well enough. P.S. In the FX6, you can disable noise reduction, while in the very similar FX3, for example, you can't. Nevertheless, there is still a slight noise reduction effect due to compression.
@Vitaphone
8 ай бұрын
Would be interesting to see your findings if it was compared to Red Raw… as it is widely touted as the only “real” raw. In my years of editing I’ve never done any scientific study as clients deliver what they deliver, but antidotally my bias has been that Red Raw is far more flexible than what you used in this comparison. Enjoyed the video.
@deadendboredom
8 ай бұрын
You echo my thoughts. The test comparison, especially including RED, has been on my wishlist for a long time, but unfortunately, I don't have a corresponding camera. Either way, RED RAW isn't the "only true" RAW - at most, it's the "true" compressed RAW. From what I've seen so far, cameras don't differ much in terms of dynamic range, especially when they're relatively current and priced a bit higher. In my view, there's often too much emphasis on brand, and there's a tendency to believe that something is better than another without actually testing it. Good lighting and the cameraman create a great image, and this often aligns with certain cameras, leading to hasty conclusions. If I do get access to a RED, that will definitely be the first test!
@antonstafeyev3606
3 ай бұрын
man... wtf... RED raw is just another compressed raw, they patented wavelet compression because it is the most efficient. The real raw only comes from ARRI - uncompressed signal, untouched.
@shueibdahir
8 ай бұрын
I have to disagree here 6:00. Blackmagic shifts its dynamic range around according to ISO. ISO is not an exposure tool in böackmagic cameras, it's used to shift the middle grey. You can choose to shift your DR towards the shadows or the highlights. On a bright sunny day i'd shoot at 800 ISO so the camera captures more of the highlights since shadows require about 3 stops in this case. During the night I'd shoot at the lowerst ISO of the second gain which is ISO 1250. That gives me 3.5 stops in the highlights and the rest is given to the shadows. That makes the camera see more into the shadows. ISO 400 and 3200 are more for balanced settings where highlights and shadows require same amount of dynamic range. The false colors can help you with this. You can check where the middle grey is and how bright it is above it or dark below it. If your shadows clip too soon, lower the ISO by 2 stops and then brighten up the image by 2 stops using the aperture, shutter angle or add lights.
@deadendboredom
8 ай бұрын
Hi, thanks for your detailed text, but there might be a bit of a misunderstanding: in my examples, I adjust the exposure afterward to demonstrate the level of detail in shadows and highlights. This is purely a technical matter and not related to achieving a "correct" exposure in terms of stops above or below middle gray. My focus is on illustrating the differences between various codecs and the pros and cons of its dynamic range.
@shueibdahir
8 ай бұрын
@@deadendboredom yeah my point was that blackmagic can essentially beat most cameras by shifting its middle grey and pull down 5 stops using the first gain at iso 800 or 4 stops with the second gain at iso 4000. You're also able to pull almost that much underexposed using the first gain at iso 100 or the second gain at iso 1250. If you use blackmagic cameras like the sony, panasonic or canon cameras you're heavily underutilising it. I wasn't talking about the correct exposure either.
@deadendboredom
8 ай бұрын
You need to explain that in more detail, preferably with a specific example and footage. At the moment, I can't comprehend why Blackmagic should outperform most cameras, and I haven't seen any practical tests supporting this claim. P.S. Are you on Instagram? Perhaps it would be more convenient to continue the conversation there in German?
@gabrielgreuter
10 ай бұрын
Great video, thanks again
@CartyCantDance
9 ай бұрын
I’m guessing before watching that compressed raw will have only marginal benefits compared to All-I with the exception of ARRI RAW. My thoughts after watching: Oh I was right, now I wish I had the lumix s5iix instead of s5ii so I could shoot ALL-I. I could always record BRAW externally, so it’s not that big a deal.
@deadendboredom
9 ай бұрын
Thank you for your before-and-after assessment! In terms of color correction possibilities and dynamic range, I wouldn't expect a significant advantage from the S5ii in All-I over the non-X version - although there would certainly be benefits in editing performance. Whether BRAW behaves in the S5ii X as it does in a Blackmagic camera regarding highlight details would need testing. As seen in the test, using a specific RAW format doesn't guarantee access to highlight recovery. With the DNG from the old Blackmagic camera, it was possible, while with DNG and Sony, it wasn't. Weren’t you a little shocked by ARRI RAW? I know i was - I invested a considerable amount of money to get it running, only to realize that, apart from more details, it seemingly offers no substantial advantages.
@ctcwired
5 ай бұрын
Curious how you corrected WB on the non-raw shots. Was it using the Chromatic Adaptation OFX? Thanks, Great video.
@deadendboredom
5 ай бұрын
For the white balance and color matching of non-RAW clips, I used the built-in Color Match function in Davinci, which utilizes the Color Checker to bring each color patch to a consistent baseline value. For me, this seemed to be the best way to not only adjust the white balance but also to get the colors as closely matched as possible. Making adjustments solely based on visuals would likely result in larger discrepancies, not accurately representing which colors can or cannot be restored. By the way, the color fringing with strong white balance adjustments in All-I clips is always present, regardless of how you adjust the white balance. P.S. With the All-I clips from the FX6, it's even possible to adjust the white balance in the RAW tab as if it were a RAW clip.
@ctcwired
5 ай бұрын
@@deadendboredom Ahh okay makes sense. I'm not sure if the Color Match function just does linearization and 3x3 matrix or if it has other fancy stuff going on under the hood. I should play with that feature more. Yeah it's cool the RAW tab works on Sony non-raw clips. They must include the necessary metadata to pull it off in their files which is cool of them. Wish more manufacturers did this. The ability to accurately adjust white balance on RAW (or any image really) just comes down to having the necessary 3x3 matrices to adjust linear gains in camera observer native RGB space based on how it acts under known illuminants (not SGamut3.cine, but wherever the bayer spectral filters land in XYZ). The main reason RAW formats feels so magical is simply because it forces users through the manufacturer's SDK where they've implemented that, when really if you use the right tooling with log footage it's all the same. In that regard, It's refreshing to see someone do a camera comparison on KZitem using ACES, so thanks for that. :p I think the artifacts shown in your video mostly just come down to flaws in the compression codec being pushed that hard which is interesting to see. Wavelet compression vs DCT, whether the debayer sharpening algorithm occurs before or after the WB adjustment, etc. Slightly unrelated, but I've had strange results with the free ProRes RAW cDNG conversion utility that's around in various app stores. Had much more accurate results with Assimilate Scratch (or just Adobe Media Converter / FCP for going to regular ProRes). I think that free utility is messing up or ignoring a colorspace conversion somewhere, at least with the cameras I've tested.
@deadendboredom
5 ай бұрын
@@ctcwired Yes, the pixel blocks definitely come from the All-I compression, which I've addressed in several videos before. But what's also interesting is how much color information gets lost in it. In one shot from the boat, for example, the yellow edge of the concrete pillar or the red railing of the ship is only properly visible in RAW. I assume that's where the low bitrate All-I compression meets the 422 colors and makes the colors disappear in small details.Do you know the name of the cDNG conversion tool? I use RawConvertor, and initially, I also had some strange deviations, but it ultimately turned out to be a bug in Resolve and that I didn't set the material up correctly.
@ctcwired
5 ай бұрын
@@deadendboredom Yeah it was “RawConverter”. Kept noticing the colors would look off in Resolve compared to FCP, or just XAVC clips. Tried doing the conversion with a trial of Assimilate Play Pro instead and then everything matched. So I think RawConverter is doing something wrong to the metadata. I didn’t want to buy Assimiliate though so these days I just use Adobe Media Converter or FCP to debayer to ProRes 4444 if I have to.
@elvisripley
10 ай бұрын
What project did that green screen come from?
@deadendboredom
10 ай бұрын
The footage is from a lighting test shoot for a concept trailer of a '60s Sci-Fi series, a project we dream of bringing to life someday.
@GlobalShutterNY
9 ай бұрын
How would Sony Long GOP compare - it seems the Long GOP may actually be better than All Intra Codecs as the trade off between compression and data rate makes the Long GOP a higher quality image (though harder for computers to edit speed wise...)??
@deadendboredom
9 ай бұрын
Based on my previous tests, where I also paid some attention to color comparison, it seemed that Long GOP actually had no disadvantage. However, this would ideally require its own video, especially considering we also grapple with the 10-bit vs. 8-bit topic here.
@GlobalShutterNY
9 ай бұрын
I am speaking specifically about cameras like the FX30 which use 10-bit internal 4:2:2 Long GOP - no question that 8-bit is not at all flexible for any sort of serious color grading... @@deadendboredom
@ludovicavice3496
9 ай бұрын
Definitely saw that the xavc i codec was super solid when trying to grade things out and its a good thing as you can fix Sony weird colors while not spending a fortune on disk space 😅 i kept reading about nikon z9 raw claims being closer to arri, better this, better that but i cant see any of these... Can you plz test them out ? 😊
@deadendboredom
9 ай бұрын
Perhaps I should embark on a deep dive into that…
@deadendboredom
9 ай бұрын
I'm here working on tests and a script, and I find it extremely challenging to delve into the aspects of look and color science. The difficulty lies in the fact that a lot of this is subjective, and there's limited objectivity in measurement. What's your feeling on the matter?
@TheLastRealJEDI
6 ай бұрын
Really good comparison. Thanks a lot!
@skymakai
10 ай бұрын
Regarding B-Raw “highlight recovery”, isn’t it false that “there’s information there”? Resolve calculates which color channels were clipped and ‘simulates’ highlight color information… right?
@deadendboredom
10 ай бұрын
Fascinating! I'm not exactly sure how it works, but even if it does like you say, wouldn't it still, in some way, be the "restoration of information"? Or at least utilizing information of one kind to restore information of another kind.
@LaurentFranchet
2 ай бұрын
@@deadendboredom The information is certainly not simulated . The data has been captured by the sensor but it is clipped when it goes through the standard decoding pipeline. Checking the highlight recovery box just ensures that this data is not clipped by processing it through a different decoding pipeline, although BM warns that unusual color artifacts may occur.
@Altcine
10 ай бұрын
Ok before I watch the video I'm going to say there's not going to be much difference between the shots from various different cams. There will be things that look slightly better to a trained eye but mostly I don't think we will see much. The times I e used ProRes Raw were amazing at bring back blown out skies etc and CDNG was great for shadows but I just shoot all-i now and am able to get so much out of it. Let's see if I regret this comment 😅
@Altcine
10 ай бұрын
Thank goodness I wasn't too far off your findings as they were very sim5to what I've found using various cams and formats. Thank you for a great video.
@deadendboredom
10 ай бұрын
@@Altcine Great that you took the plunge! Best of luck for your work!
Пікірлер: 46