-
SONY EX-1 Footage Final Render 8Bit or 10Bit?
-
Wow, thanx for the vote of confidence! 🙂 I guess, I never saw it from that angle.
Very interesting point.Sascha
-
[Sascha Engel] “I guess, I never saw it from that angle.
Very interesting point.”They call it wisdom Sascha… It’s about the only benefit there is to aging I can think of.
You’ll be thinking of me twenty years from me as you step up to the podium after they play that retrospective. 🙂
David Roth Weiss
ProMax Systems
Burbank
DRW@ProMax.com
http://www.ProMax.com
Sales | Integration | SupportDavid is a Creative COW contributing editor and a forum host of the Apple Final Cut Pro forum.
-
Now, I am seriously blushing. Well, if that ever happens, you have my word on the COW:
I will invite you as guest of honor for that ceremony!Thank you!
Meanwhile I will see if I can get some place on my HD, since our salaries here in the Middle East are just not cut out for constant HD buying.But wow, that posts of yours certainly made my day:-)
Sascha
-
If someone’s looking at a 10th gen copy of this work it’s going to be derived from the HDCAM deliverable and I’d say that the HDCAM tape (1440×1080, 3:1:1 8-bit, interlaced) is a already a step down from the source ProRes LT footage (full raster, 4:2:2, 10-bit, progressive). And for pieces where a master QT exists, a 100th gen copy of the file is going to look exactly like the master QT (assuming no data corruption happens during one of the copies). It’s not like BetaSP where every analog step away from the master tape is a notable step down in quality. Sure, if the generation pipeline looks something like QT Master->HDCAM SR->HDCAM->DVCProHD->Blu-ray->Torrent->YouTube->DVD then the end product will look far worse than the Master. But at that point it doesn’t matter if the Master was PR LT or PR Regular.
We all have practical limitations, like budgets, and the trick is to do the best you can with what you have (I hear 5k RAW is sweet, but how many have the means to handle it cradle to grave?). At the old gig there wasn’t a budget to build two new 100TB Xsans (one mirroring the other like they have now) so the option was either staying in DVCPro HD or moving to ProRes LT (better quality at about the same file size). I guess turning away work would’ve been a way to decrease our data footprint and allow us to use ProRes regular but that just sounds like a bad idea to me. 😉
I even walked the exact same footage in ProRes regular and ProRes LT through our post cycle (editing, rendering, exporting, tape layoff, web compression, etc.,) during the evaluation process and saw no difference between the two in our workflow. In other situations the results might not be the same, but in our workflow they were. Trying to get a 6 figure corporate expenditure approved for little to no gain just didn’t seem worth it. Using those funds to buy a bunch of Ki Pros so we didn’t have to use Cineform or XDCAM EX/HD for field capture, now that was money well spent. Sure, if someone is being stingy over a few hundred, or even a couple thousand bucks that can be short sighted, but in the situation I’m talking about it would’ve been hundreds of thousands of dollars to sufficiently upgrade the storage.
To further derail this thread, why do you keep equating PR LT w/tossing away half your data (especially when the sources being discussed are of lesser quality than PR LT)? PR LT, like PR regular, is full raster 10-bit, 4:2:2 it’s just about 30% smaller in data rate. Sure, 30% could be a difference maker but when the source footage is 4:2:0, 8-bit and nearly 1/4th the data rate of PR LT I’m not immediately convinced that bigger is better. That’s not to say I wouldn’t choose regular PR just to be safe (I cut a DV-sourced doc in a PR timeline and did PR exports) but there’s certainly a point of diminishing returns. I mean, I doubt you work exclusively in Uncompressed even though, on paper, it should provide a better quality master than any flavor of ProRes or DNxHD. 😉
-Andrew
2.9 GHz 8-core (4,1), FCP 7.0.3, 10.6.6
Blackmagic Multibridge Eclipse (7.9.5) -
I was writing a long shitty post but i’ve deleted it.
In short:
– Compression means degradation (even if your eyes can’t see it).
– The only reason for compression is economy (money).
– Your source shouldn’t determine your workflow. A poor quality footage doesn’t justify a poor quality workflow. You can decide a lesser quality workflow for speed, HD space, system performance or whatever other constrain you may find reasonable, but never because your footage is not the best. In fact i see this the other side: The poorer the footage, the more care deserves.
– The same about your final delivery format. Whatever the delivery format will be, my target is a master with the higher quality i can get from my means.Just a technical add about this case:
When you put an 8b compressed file in FC sequence and you apply an effect like MagicBullet (that works in 10b) and you set your sequence for 10b High Precision. the stuff in your sequence is not anymore 8b compressed, but full 10b, with 1024 luma values and 1024 Cb/Cr values.I don’t dismiss ProresLT (I haven’t used not a single time).
I think is a great option for 10b massive data storage and high quality distribution (quality/data-rate), but is not a post-production codec. Not even Apple recommend it for post (see the Prores White Paper). LT is not ideal for stuff that will undergo further heavy processes.Then you may or may not see the difference on an screen (you’ll hardly see it in YouTube), but the difference exists.
Is a pity that nobody has yet made a good comparative test between the different Prores flavors. I mean a test as these made years ago by Marco Solorio (https://www.onerivermedia.com/codecs) with the QT codecs available at the time.
To judge a codec for postproduction you don’t use a monitors/scopes (that’s good for judging codecs for distribution), because is not just about what your eyes see, but about what your eyes do not see.
Is about what the machine see.
If you have time, please have a look to the link I’ve posted and click in the “Tests” tab. Is quite interesting to see how a proper test should be done.
rafael -
Thanx for that long and detailed answer. Of my current project I already re-batched everything in ProRes instead LT. About the Rock Concert, I will just have the client to be patient and wait – I have to re-render everything set to Apple ProRes 10Bit Hi Precision – which will take ages on my system, but yes, if there’s a difference it’s worth the pain…..Jumping back to work, clock is ticking against me….leaving in 7 days to Thailand :-).
Very nice test, by the way!
Sascha
-
Thailand?
Great. This year the Sonkram festival is from 13 to 16 of April.
Enjoy it. -
Hi Rafa,
will write you a mail soon…wanna know where that Festival is …
But not here, that is a little off topic 😉 Keep work and travel separated!Sascha
-
I’m glad you decided to do away w/the shitty post ’cause I don’t really see the reason to write one.
Basically, I agree w/everything you said. In a situation where there is an unlimited budget, no deadline and no technological limitations there shouldn’t be a reason to work at anything less than uncompressed 4:4:4 for everything.
I’ve never had a chance to work on a project like that so I’ve got to find ways to do the best I can given the budget, time and tech limitations of whatever job I’m working with. What I find a bit bewildering is the attitude you and David are taking towards me w/o ever trying to learn more about the situation. It’s like saying I made an inferior choice by choosing a bike for particular job instead of a car w/o everything asking about the details of the job and why I thought a bike was a better choice.
[Rafael Amador] ” Not even Apple recommend it for post (see the Prores White Paper). LT is not ideal for stuff that will undergo further heavy processes.”
From page 5 of the PR White Paper (which I am familiar with:
“Apple ProRes 422 (LT): Like Apple ProRes 422 (HQ) and Apple ProRes 422, the new Apple ProRes 422 (LT) codec supports full-width 10-bit video sequences, but at a target data rate even lower than these siblings. Apple ProRes 422 (LT) weighs in at 100 Mbps or less, depending on the particular video format. It balances incredible image quality with small file sizes, and is perfect for digital broadcast environments where storage capacity and bandwidth are often at a premium.
Apple ProRes 422 (LT) is ideal for live multi-camera and on-location productions where large amounts of footage are acquired to disk. The low data rate of Apple ProRes 422 (LT) also makes it an excellent choice for transcoding complex camera codecs like AVCHD.”Every thing Apple says PR LT is good for describes part of the workflow at the last job I had. A lot of live event coverage with same day edits where getting your coverage up before your competitors is paramount. Workflows where getting a 10min export down to 7min is a big deal, where TBs of footage are being generated (at ProRes LT settings) on location and external HDDs are shuttling between locations so that everything ends up in the centralized storage ASAP.
I agree that PR LT isn’t good for heavy processing in post and even stated so earlier in this thread. Sure, if you are doing a lot of manipulation in post PR LT isn’t going to hold up but if you are doing little to no manipulation is stays together just fine in my experience. That doesn’t mean you shouldn’t test it in your own workflows, but in the workflow I tested it in it worked just fine. It exceed the quality of the previous workflow (DVCPro HD based) while not exceeding the storage and time limitations that were beyond our control.
I would love it too if someone regularly did comprehensive codec round ups. And not just codec roundups but compared different NLEs and compression software to see how they each handled a variety of codecs in common transcoding scenarios.
Like I said before, I always strive to maintain the highest quality I can given the limitations of the job. I’ve worked on prime time shows where no one batted an eye at renting two 40TB Avid ISIS’s for redundancy and I’ve worked on low budget documentaries where post was held together by superglue and bailing wire. It both cases I helped setup and/or maintain the highest quality possible given the situation.
In a perfect world I’d maintain a perfect workflow but we don’t live in a perfect world. 😉
-Andrew
2.9 GHz 8-core (4,1), FCP 7.0.3, 10.6.6
Blackmagic Multibridge Eclipse (7.9.5) -
Andrew,
I fully agree with you. As I said, compression is a matter of economy and as you said: In a situation where there is an unlimited budget, no deadline and no technological limitations there shouldn’t be a reason to work at anything less than uncompressed 4:4:4 for everything.I’ve insisted just because I know Sascha has not those constrains to force him to work with LT.
BTW, the recomendation by Apple of using LT to transcode AVCHD is pure marketing shit.
Cheers.
rafael
Reply to this Discussion! Login or Sign Up