|
|||||||||
|
Thread Tools | Search this Thread |
August 19th, 2009, 10:25 AM | #1 |
Tourist
Join Date: Aug 2009
Location: Kista, Sweden
Posts: 3
|
Testing AVCHD encoders - standardized AVCHD test clip?
Assume that I have footage in AVCHD that I need to edit and where I *must* deliver the final result as AVCHD.
If I understand messages in this forum right (?), I can achieve superior final quality by transcoding to a less compressed format such as Cineform first, then editing, and finally recompressing to AVCHD, rather than using a native AVCHD editor throughout the workflow. But this must mean that the Cineform-to-AVCHD encoder is higher quality than the native AVCHD editor's AVCHD encoder (because if the latter had used an internal intermediate format similar to Cineform, it should have been able to achieve a similar result). A conclusion is that there must be very large differences in AVCHD-encoder quality between different editors. I'd be very interested in seeing some kind of objective evaluation of the quality of AVCHD encoders. I guess this could be accomplished by comparing the results of re-encoding a short, standardized AVCHD test clip. I humbly would find it great if the experienced members of this forum agreed on a representative test clip that suitably exercises AVCHD encoders. Already done, perhaps? Or is the stage set for a competition or a vote? |
August 19th, 2009, 11:02 AM | #2 |
Inner Circle
Join Date: Nov 2006
Location: Tallahassee, FL
Posts: 4,100
|
You are drawing an incorrect conclusion from a faulty premise.
Your premise is that because of what you've read here, Cineform as an intermmediate codec must produce superior results to just using AVCHD on the timeline. This is incorrect. If I place File A on the timeline and simply render to File B, that is going to give me the best results If I place File A on the timeline, render to cineform to get file x, then load file B on the timeline and render to file Z, then File Z will not be as good as File B. HOWEVER, if I put file A on the timeline, make some cuts, gange the sky to be more blue, make the grass more green, enhance the colors, increse the contrast, and THEN render the file, then the cineform workflow should produce superior results.
__________________
DVX100, PMW-EX1, Canon 550D, FigRig, Dell Octocore, Avid MC4/5, MB Looks, RedCineX, Matrox MX02 mini, GTech RAID, Edirol R-4, Senn. G2 Evo, Countryman, Moles and Lowels. |
August 19th, 2009, 02:13 PM | #3 |
Regular Crew
Join Date: May 2009
Location: Manhattan
Posts: 131
|
Somewhat confusing
Just to get things clear here.
Let's say your file A is original AVCHD footage. Let's also say you transcode that AVCHD footage into Cineform (or Apple Intermediate Codec, or ProRes) for easier editing, and call that file X. If you edit A directly in you NLE (just cuts and simple fades, wipes, etc) and render the resulting content to file B in AVCHD (directly from your NLE), you will preserver the highest amount of quality of the original AVCHD. If you edit X and then render to B in AVCHD, the outcome will theoretically be inferior then A -> B, because you now have A -> X -> B. However, if you need to do some significant post-production work on your video (colour correction, exposure correction, visual effects, etc), the better solution would be A -> X -> B. This is becausse AVCHD has 4:2:0 chroma sampling, whereas all the popular intermediate codecs (ProRes on Mac, or Cineform on Mac/Win) have 4:2:2 chroma sampling, giving you much better colour resolution to work with. The original colour won't automatically be resampled to the higher resolution, but any image manipulation will gain significantly from the additional colour space. When your edit is finished, your NLE will down-sample your chroma data back to 4:2:0 for AVCHD encoding. So, choose one or the other depending on what type of editing work you need to do to your AVCHD, and how much of it will there be. If it's just transitions, edit AVCHD directly. If it's whole clips (or whole movie), transcoding to something intermediate is a much better idea. |
August 19th, 2009, 05:59 PM | #4 |
Inner Circle
Join Date: Nov 2006
Location: Tallahassee, FL
Posts: 4,100
|
Pedrag explained it more clearly than mine. I was seriously sleep deprived. There is one things left out.
Cineform and some other intermediate codecs can also be 10 bit of precision instead of the 8 bits of Normal ProRes, AVCHD, etc. This allows for much cleaner color gradations, luma changes, and other things. So it's highly desirable to use a 10-bit codec for that kind of work if you can. It does require more machine power though.
__________________
DVX100, PMW-EX1, Canon 550D, FigRig, Dell Octocore, Avid MC4/5, MB Looks, RedCineX, Matrox MX02 mini, GTech RAID, Edirol R-4, Senn. G2 Evo, Countryman, Moles and Lowels. |
August 19th, 2009, 06:42 PM | #5 |
Inner Circle
Join Date: Mar 2003
Location: Ottawa, Ontario, Canada
Posts: 4,222
|
IF you just want to cut off shots of one's feet etc!!! Then the easiest is to edit in camera by splitting clips . Absolutely no quality loss. You can still backup everything on the computer for future use. Not sure how easy this is on all AVCHD cams but is very easy on the Sony's I have and is a skill worth learning( and a lot easier than using a smart cell phone!!!!).
Ron Evans |
August 20th, 2009, 03:39 AM | #6 | |
Tourist
Join Date: Aug 2009
Location: Kista, Sweden
Posts: 3
|
Thank you, Predrag and Perrone, for your clarifications. Apologies for my naive perspective.
Quote:
In an ideal world, I would wish that the difference between native AVCHD editing and AVCHD-intermediate-AVCHD editing only be that the native NLE uses something equivalent to a high-quality intermediate codec _internally_ on rendering to screen or disk, but that this is hidden from the user. If one needs to go between several programs, the situation is different of course, and it would be necessary to export the intermediate format. |
|
August 20th, 2009, 06:50 AM | #7 |
Regular Crew
Join Date: May 2009
Location: Manhattan
Posts: 131
|
Internal rendering
I'm not sure how others do it, but Apple's software (iMovie, FCE, FCP) render an effect by adding a file that contains only the modified portion(s) of the original clips. For example, if you have a transition between two clips, a new file will be created that only contains the transition. Original files will remain the same. The new file is usually in the same format as the original ones (AIC, DV, ProRes...).
If Apple (or Clipwrap) one day makes it possible to work with AVCHD directly, I wonder how the edited clips will be rendered. What you are saying makes a whole lot of sense. If that is indeed the case, the "internal codec" may as well be some standard one (such as AIC, or ProRes) that is supported across the OS (via QuickTime in Mac, or AVI in Windows). In that case, there should be no reason why external apps couldn't work with those temporary renders. I am now very curious as to how FCP deals with various codecs thrown at it. |
August 22nd, 2009, 05:07 AM | #8 | |
Regular Crew
Join Date: Apr 2009
Location: Nelson, New Zealand
Posts: 63
|
Quote:
|
|
| ||||||
|
|