Thread: Komodo 12.2 stops of dynamic range in CineD test.

Reply to Thread
Page 9 of 11 FirstFirst ... 567891011 LastLast
Results 81 to 90 of 104
  1. #81  
    Senior Member
    Join Date
    Dec 2010
    Location
    Toronto & Vancouver
    Posts
    4,177
    Quote Originally Posted by Bj÷rn Benckert View Post
    Ok so what about shooting 8k, do signal cleaning within 4 pixels and then downsample to 4k... Now such signal has not "lost" or been treated destructivly compared to a camera that just captured 4k at sensor level. On the conterary the 8k downscaled to 4k benefit heavily with almost four times better guess work for the debayering process and the noise reduction done has not really caused any lost any info compared to the straight 4k capture, on the conterary.

    If comparing cameras you kind of need to bring them to a common ground before comparing, which is difficult so yes, it┤s hardly ever just cameras that is being compared but also the post process, or the in camera baked process versus what could be done in post or what have you.
    I guess this isn't about Komodo anymore, but the same logic applies, imo. It's more about comparing/testing against the advertised numbers in this case, anyway. Bear in mind, I'm not trying to convince any one, just noting that it'd be gnarly if you have to jump through hoops in post merely to meet the claimed specs (regardless of if NR is part of your workflow or not).

    As for your example, is it an actual 8k camera or a 4k one? No doubt there'll be (substantial) benefits with even the mildest noise reduction with such a large downscale (more so than 6k Komodo), but RED doesn't spec/advertise their cameras as '16+ stop *4k* cameras from 6k~8k sensors when post NR is applied'. Hence, whatever advantages downscaling/NR provides is still more of a bonus, in the same way that 8k allows for 4k reframing/image stabilization is a bonus (you wouldn't typically rely on it; you'd change the lens or use a gimbal/steadicam/dolly when possible).

    Forgive this exaggerated/hyperbolic anecdote to demonstrate the NR point further, but it's easier/faster to render a sky replacement on baked 4k footage than it is to use NR on 6k~8k raw footage; would we factor that in to the "usable" DR numbers of a camera that shoots baked 4k (to merely hit its advertised numbers, no less)? No, of course not; cause it's post.

    Seems most here happily accept applying post-NR as a means to get closer to the advertised specs (which is fine). I'm just more inclined not to, partly because of how much RED's cost compared to similarly specced cameras (ironically some of which include software with decent post-NR, but don't need it to hit their DR spec), partly because it looks like a bigger delta in this particular Komodo test, and partly because I've seen how these hacks/tricks can fall short with my own (non-Komodo) RED footage.

    And yes, it's difficult to have an apples to apples comparison, but I'd think adding more sensor/codec/resolution/model specific post tweaks and tricks (none of which work on all cameras in equal measure), is not the way you're supposed to go about objectively testing capture capabilities (again imo; feel free to disagree).

    Anyway, we're drifting from Komodo, which is where these concerns stemmed from (the internal Highlight Recovery that looks to be a hefty two highlight stops, which I presume/hope is quite a bit more aggressive than other RED sensors... or at least Dragon, which was designed/measured/metered pre-IPP2 HR at a claimed 16+). It seems this thread is diverging into two issues; one about NR's effect on testing for DR (even though it was brought up as a fix for pushing Komodo's ISO higher for a more even over/under), and the other being Komodo's monochrome-ish highlight recovery and how to best deal with it (so it has the least adverse impact on the useable DR).

    Unfortunately, the last couple of days/pages haven't provided any new answers/info, so we're kinda talking in circles now.
    Last edited by Mike P.; 01-26-2021 at 12:23 AM.
    Reply With Quote  
     

  2. #82  
    Senior Member
    Join Date
    Aug 2007
    Location
    Macau
    Posts
    2,524
    Komodos are being used everywhere producing broadcast and cinema content. Its selling like Hotcakes. I've used one and it intercuts almost seamlessly with my Monstro. I would assume the professional world thinks what this camera brings to the table is more than enough.

    Honestly nowadays you get to a point you have to trust the company that makes the camera. Red is a Digital Cinema Company. Just like Arri. They are on the business of making professional cameras targeted to motion pictures and high end, broadcast and comercial content. They have a name to defend, a brand to keep. Its their bread and butter. Other companies, like Sony, Panasonic, Canon etc. have various different divisions and target markets, from consumers, prosumers, to higher end markets. Each of their offerings is well defined in category and you should expect performance related to these target points.

    The Komodo however was developed as a B cam or a camera that could intercut with the other A cams well, with action in mind. Like Jarred said, something that would survive the mere seconds audiences accept from Prosumer and POV cameras in High End Productions. The internal REDCODE RAW compression in the Komodo is superior to every single offering in he price range for any other manufacturer not only in the price range, but several thousand dollars above. The global shutter is a unique feature that opens amazing creative possibilities. the dynamic Range? According to Red, close to Helium. All of this for 5000 USD. Its an incredible deal.

    I say try the camera yourself, like Bob Gundu said in the beginning.
    SÚrgio Perez

    Weapon Monstro #03294 "Amochai" in Macau

    Video Director/Creative/Producer


    http://vimeo.com/user1503556

    https://www.youtube.com/user/spzprod..._as=subscriber
    Reply With Quote  
     

  3. #83  
    Quote Originally Posted by Mike P. View Post
    Anyway, we're drifting from Komodo, which is where these concerns stemmed from (the internal Highlight Recovery that looks to be a hefty two highlight stops, which I presume/hope is quite a bit more aggressive than other RED sensors... or at least Dragon, which was designed/measured/metered pre-IPP2 HR at a claimed 16+).
    2 stops is not unusual from my experience. Usually you cannot get the data pre-recovery to know in most cameras. Every now and again you get one that out puts DNG files and then you can see everything pre-recovery.

    I don't have a step chart here from my Epic W but it would be interesting to see how it compares, that's true. Pushing the WB and then seeing how the channels separate is a possible way. But i was thinking about this. If the WB was done after HR then the reconstructed channels would separate too - so maybe some cameras do it that way...

    HR in the real world may yield different results depending on scene.

    So when the Sony FS700 first did RAW via the Odyssey 7Q it was uncompressed DNG and i spent a while ripping them apart. I discovered the green channel was about two stops more sensitive than the others which meant that HR played a huge role in that cameras range (which was perhaps 11/12 stops overall). Since then i've been quite aware that HR has always played a big part in the images we use. I realised that all our still photos in Lightroom are all using HR as well and you cannot turn that off.

    So if you have a camera producing baked images then NR is most likely in play and maybe HR as well.

    So as you say it's very difficult to compare. One scene could look perfect with HR and another scene might not look as good.

    Also the HR approach doesn't have to be the copy one channel from one to another, it could be more nuanced and take the surrounding image into account and then in some cases you'd never know. Thinking skies, skin and so on.

    Hence saying that all these ranges values are 'valid' you just need to know what you're talking about - step chart range, colour range, HR range...

    Bearing in mind most of us are shooting people and often indoors with the world through windows - for me that's the test. If i can light 'normally' inside and still see outside (with a bit of a grade) then that's all i am looking for. My EpicW does this, and hopefully Komodo too. Of course if you are underwater or doing specialised work then other factors come into play - colour crosstalk and maybe colour range. YMMV

    cheers
    Paul
    -------------------
    insta: @paul.inventome
    Reply With Quote  
     

  4. #84  
    Senior Member Christoffer Glans's Avatar
    Join Date
    Jun 2007
    Location
    Stockholm, Sweden
    Posts
    4,784
    Quote Originally Posted by Mike P. View Post
    Seems most here happily accept applying post-NR as a means to get closer to the advertised specs (which is fine).
    Advertised specs refer to the "by eye" visible stops, meaning, yes, it has 16,5 stops. If Red claims Komodo has 16,5+ stops in any technical measurable way, there's no test really showing that number in any usable ways. However, 14 stops are actually there and able to be used.

    But the core thing still remains; we have two types of cameras in the market, RAW systems that use post-processing in post to achieve results with much greater control and consistency, and in-camera processing systems that do as much as possible before condensing the image down to a finalized image directly out of the camera.

    The different systems have different pros and cons and I've argued many times that the distinction between "video camera" and "cinema camera" has new meanings and relevance today. A video camera system focuses on fast turn arounds with lots of built-in tech that makes it easier to plug and play very fast and get a good result with minimal post manipulation. A cinema camera system focuses heavily on interaction with other equipment to achieve a perfected result by having lots of control throughout the entire process based on slower and more methodical approach to both image quality and sync with VFX, motion graphics and sound design.

    So how do you really measure advertised specs for a cinema camera system that only gets full potential after the entire pipeline is finished? There's no heavy production ever not doing heavy manipulation of the camera material in post and the potential is reflected in the possible outcomes rather than the input.

    Either you measure RAW cinema camera systems and compare between only them and not in-camera processed video systems, or you put the cameras through the entire pipeline to measure the end quality. There's no other way of measuring systems. Putting all systems into one big pile trying to standardize testing for everything ignores the reality of how these systems work. It's ignoring the intention and differences based on nothing but the current idea of only pixel peeping the direct out of camera images.

    I've rarely seen any test that actually focuses on what we actually do with these cameras. The best example of a good test that does things totally different than everyone else is Steve Yedlin's test where he put everything though his post-process method. It's much more showing of the quality of each camera (even though he makes a point of not telling which is which) than any attempt at measuring the RAW image data directly out of the camera.

    I think that people need to re-invent how we test these systems. First, ditch the big pile and divide them into video and cinema groups. You can compare end results in different types of work, but not the core systems. Test the RAW capabilities like Gunther here has done, but combine it with a standardized test of production. Not charts, not people in standard bad film school lighting, but having a standardized set that's locked and that shows low key, high key, models of different skin types, high contrast sun, low contrast soft light, smoke etc. Then putting all of those shots into a standardized end goal of 4K DCI mastering like everyone is doing with these cameras.

    That way we would both see the RAW data, but also how well the camera actually behaves in real-world situations. The combined result of such tests gives a true overview of all systems and just like Steve Yedlin's test, would be much more telling of which system has the most potential. Combining that result with the physical specs of the camera, you can derive a point score based on RAW image quality, post-processing range, end result limitations and physical possibilities.

    Without having that test done on Komodo, we could speculate what it might get. Mid-quality score on RAW image quality, mid-quality on post-processing range, mid-high-quality on end result and high-quality on physical possibilities. While a camera like Monstro gets high-quality on RAW image quality, super-high-quality on post-processing range, high-quality on end result and mid-quality on physical possibilities. Speculative, but such a standardized test would seriously rank cameras in a much better fashion than the sum of all the half-measured tests that everyone is doing everywhere right now.
    "Using any digital cinema camera today is like sending your 35mm rolls to a standard lab. -Using a Red is like owning a dark room."
    Red Weapon 6K #00600 Red Komodo #002397
    Reply With Quote  
     

  5. #85  
    Senior Member
    Join Date
    Dec 2010
    Location
    Toronto & Vancouver
    Posts
    4,177
    Quote Originally Posted by paulcurtis View Post
    So as you say it's very difficult to compare. One scene could look perfect with HR and another scene might not look as good.
    ...
    Hence saying that all these ranges values are 'valid' you just need to know what you're talking about - step chart range, colour range, HR range...
    Admittedly that's the kind of guess work I'm talking about/prefer to avoid, and why I'm not quick to write-off concern(s)... Where there's no good way to know for sure if it's going to look good/have rgb in a (particularly meaningful) highlight or is going to look gaudy/mono-ish/unrecoverable/only good for roll-off. Most other platforms don't rely on highlight recovery stops in there even over/under numbers/metering, and in my experience that works better even if they're "only" 12stops at +6/-6 (whereas CineD's Komodo scopes look to be sitting at -8/+4, unless you compensate to 3200ISO with NR).

    (FYI: I could be wrong, too. Nick mentioned earlier that the mids are noticeably cleaner on Komodo... Maybe it's enough to bridge the over/under gap without pushing the ISO too high, despite CineD's scopes and sample scene pointing to the contrary. In that case, it'd be easier to compensate for and allievate concerns.)

    Quote Originally Posted by Christoffer Glans View Post
    Advertised specs refer to the "by eye" visible stops, meaning, yes, it has 16,5 stops. If Red claims Komodo has 16,5+ stops in any technical measurable way, there's no test really showing that number in any usable ways. However, 14 stops are actually there and able to be used.

    But the core thing still remains; we have two types of cameras in the market, RAW systems that use post-processing in post to achieve results with much greater control and consistency, and in-camera processing systems that do as much as possible before condensing the image down to a finalized image directly out of the camera.
    I disagree if only because it isn't the case for pretty much every other camera outside of RED. Take P6k's (Komodo competitor) or ALEV3's (DSMC2 competitor) scopes and advertised numbers; they have definable rgb in the same amount of stops as their advertised specs (~14 for Alexa and ~13 for BMD)... the "+" is the 1~2 iffy half-clipped highlight stops and 1~2 shadow stops in the noise floor beyond those full numbers (and could be recovered to varying degrees/quality). And that's ignoring that those competitors have an even over/under at base ISO (aka a useably clean ISO without NR).

    As for categorizing raw vs baked/processed capturing, the distinction is made inherently by the camera, y'know, capturing raw or baked codec. The baked codec will invariably have less to play with, but even if their using NR and HR internally, they can't make-up info that wasn't there on the sensor (only maximize it at the cost of having even less room/colour/info to play with in post). To my knowledge, even with whatever internal processes they're using (whether that's Alexa's prores or mirrorless's 420 h264/h265), I don't think the gains are as big as Komodo's +2 mono-highs and +2 noisy-lows claims (which is a full 25% of its stated specs and, again, wouldn't even the over/under if you choose to count them anyway)...

    ...It's easy to check too; just compare internal codec vs. external raw of a Sony or Panny or Nikon mirrorless or the R5's 8k raw vs 4kHQ mode (which would also have the advantage of internal downscale)... I'm guessing it'd be less than a stop difference between the raw and codec captures (in terms of noise... the raw probably has more useable DR *without* processing). And if that is the case, the "but codecs use internal processing and raw doesn't" logic really won't hold.
    Last edited by Mike P.; 01-27-2021 at 01:37 AM.
    Reply With Quote  
     

  6. #86  
    For me the main issue is that RED rates Komodo at 16+ stops!

    I am sure if you don't worry about color info clipping and also are willing to accept really high noise, you can get "16+" stops.

    But it comes across like cheap marketing hype that in turn cheapens the brand.

    Bruce Allen
    www.bruceallen.tv
    Reply With Quote  
     

  7. #87  
    Senior Member Jacek Zakowicz's Avatar
    Join Date
    Jun 2008
    Posts
    3,522
    I'm just surprised that after 15 years people are still up in arms about RED marketing...
    Jacek Zakowicz, Optitek-dot-org, jacek2@optitek.org
    Professional Broadcast and Digital Cinema Equipment
    Reply With Quote  
     

  8. #88  
    Quote Originally Posted by Jacek Zakowicz View Post
    I'm just surprised that after 15 years people are still up in arms about RED marketing...
    I'm not. I just mention it because people may wonder why I'm on the forum but rent Alexa instead, because I prefer to rent cameras made by a camera company that lies less.

    Don't get me wrong, I like RED. RED's just like your one buddy whom you like a lot but can't their hype take seriously all the time.

    Bruce Allen
    www.bruceallen.tv
    Reply With Quote  
     

  9. #89  
    Senior Member
    Join Date
    Dec 2010
    Location
    Toronto & Vancouver
    Posts
    4,177
    LOOOL! I'm out here writing paragraphs to explain my hesitation/reservations, then Bruce and Jacek basically hit the nail from both sides in like 4 sentences. Ha!
    Reply With Quote  
     

  10. #90  
    Senior Member Christoffer Glans's Avatar
    Join Date
    Jun 2007
    Location
    Stockholm, Sweden
    Posts
    4,784
    Quote Originally Posted by Mike P. View Post
    I disagree if only because it isn't the case for pretty much every other camera outside of RED. Take P6k's (Komodo competitor) or ALEV3's (DSMC2 competitor) scopes and advertised numbers; they have definable rgb in the same amount of stops as their advertised specs (~14 for Alexa and ~13 for BMD)... the "+" is the 1~2 iffy half-clipped highlight stops and 1~2 shadow stops in the noise floor beyond those full numbers (and could be recovered to varying degrees/quality). And that's ignoring that those competitors have an even over/under at base ISO (aka a useably clean ISO without NR).
    But it is the case. My argument was general and did not have to do with Red specifically at all. I would want all RAW systems to be separated from in-camera processed systems. And the numbers they give might be different from Red because Red seems to go by "per eye" viewpoint of the Xyla chart and do their numbers from that, while Arri might as well do it more technically measured and not counting what we see with our eyes on the chart. It doesn't really matter really because my point is doing all RAW systems equally and ditching all systems of in-camera processing to measure with.

    Quote Originally Posted by Mike P. View Post
    As for categorizing raw vs baked/processed capturing, the distinction is made inherently by the camera, y'know, capturing raw or baked codec. The baked codec will invariably have less to play with, but even if their using NR and HR internally, they can't make-up info that wasn't there on the sensor (only maximize it at the cost of having even less room/colour/info to play with in post). To my knowledge, even with whatever internal processes they're using (whether that's Alexa's prores or mirrorless's 420 h264/h265), I don't think the gains are as big as Komodo's +2 mono-highs and +2 noisy-lows claims (which is a full 25% of its stated specs and, again, wouldn't even the over/under if you choose to count them anyway)...
    Any processing will improve on the measurement of stops. It's even stated in the test "manual" that this is the case for noise reduction on footage etc. I just provided a processed chart of Xyla showing how much to gain from processing and cameras using in-camera processing already do this. Also, doing this internally can sometimes be powerful since it's done before any codec coding is achievied. Only advantage Red has is that R3D is pretty much the best cinema codec in existence. As I've mentioned in other threads as well, R3D is the true power of Red, not their cameras or anything of the sort, R3D is where Red has power in the industry.

    Quote Originally Posted by Mike P. View Post
    ..It's easy to check too; just compare internal codec vs. external raw of a Sony or Panny or Nikon mirrorless or the R5's 8k raw vs 4kHQ mode (which would also have the advantage of internal downscale)... I'm guessing it'd be less than a stop difference between the raw and codec captures (in terms of noise... the raw probably has more useable DR *without* processing). And if that is the case, the "but codecs use internal processing and raw doesn't" logic really won't hold.
    External RAW is external RAW, not internal. It's the same as how Apple use hardware and software in tandem. Install Mac OS on a PC and it "can" run, but it's unstable and sometimes not working at all. Remember one system like ten years ago that I made into a Hackintosh that literally burned the GPU processer to the point I saw purple smoke through the fan vent. If comparing RAW vs codecs, the codec first needs to be internally processed and as low in destruction as possible. Most systems mentioned use super compressed codecs where there's no point in having a good starting point because you can't do anything at all with the footage before it breaks down. And RAW is crippled by other factors like the camera overheating and affecting black levels.
    In the end it becomes a clusterfuck of reasons why some systems aren't going to work. Compare systems that DO work in RAW and judge the post handling. In the end, if the codec or RAW files breaks down, then "more stops" doesn't really mean shit compared to high quality RAW that gets processed.

    I would love to see actual comparisons between Arri/Red systems and mirrorless/DSLR systems that do RAW. I mean, so many companies use "RAW" that I think people forget that companies like BM actually do in-camera processing before their files go into their "RAW" file format, since otherwise they have to pay Red for compressed RAW legally. Only true uncompressed RAW is considered RAW, otherwise, it's licensed from Red as compressed RAW.

    So in my opinion it's still the case that true RAW systems should be kept separate from everything else and judged in tandem with processing and final output as I described in my previous post. It's the only way to judge cinema cameras and end result is also the only way to compare against other systems.

    It's actually perfectly justified to do tests as Steve Yedlin did; take all cameras and do post-processing and judge how they behave. His test is a bit old now, but I would have loved to see him doing an update with, 65mm, Alexa 65, LF, Monstro, Helium, Regular Alexa, Komodo, Gemini, C500, C70, FX6, Venice and all valid DSLRs to see how they all perform in a real processed workflow... because every one of us does a processed workflow essentially, so such a test is more accurate than anything else.
    "Using any digital cinema camera today is like sending your 35mm rolls to a standard lab. -Using a Red is like owning a dark room."
    Red Weapon 6K #00600 Red Komodo #002397
    Reply With Quote  
     

Tags for this Thread

View Tag Cloud

Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts