Best blu ray compression method for grainy black & white film
Hello everyone, thanks so much for reading. I'm facing the challenge of getting the best possible compression for a 90 minute B&W feature film for blu ray. This blu ray will be used for exhibition in a large theater so I want it to look as good as is technically possible. I'm burning to a 25 GB blu ray disc.
The original format is grainy B&W super16mm which was then telecined to Pro Res 422 (HQ) which is my source codec. My Pro Res files look fantastic and I've had a great-looking DCP made from them. But now that I'm encoding the film for blu ray, the excessive grain is translating into pixelation/artifacts in the encode. I do realize that I probably won't be able to completely eliminate compression artifacts, but I want to get the very best result I can and would love to hear any suggestions that anyone might have.
My system is a Mac running 10.8.5, 2.7 GHz Intel Core i7. My source codec is Apple Pro Res 422 (HQ).
I am encoding with adobe media encoder CC 18.104.22.168 64 bit and/or Compressor 4 version 4.0.7.
Here are the tests that I've done so far:
Using compressor, I generated an "H.264 for Blu-ray" file that came out to 17.85 gb. I chose multi pass and maxed out the average and maximum bit rates at 30 and 35. Here are my settings in detail:
Description: H.264 elementary stream for Blu-ray and AVCHD authoring
Width and Height: Automatic
Selected: 1920 x 1080
Pixel aspect ratio: Square
Padding: Preserve source aspect ratio
Frame rate: (100% of source)
Stream usage: Blu-ray
Average bit rate automatic,
selected: 30 (Mbps)
Maximum bit rate: 35 (Mbps)
Next, using Adobe Media Encoder, I created both MPEG2 and H264 files.
For the MPEG2 in Adobe Media Encoder, I generated a file that was about 21 GB, (which is probably the largest file size I can fit on a 25 gb blu ray with my two PCM audio streams). I noticed that in the particularly difficult to compress areas (which are those that are both grainy and out of focus in the original film) the MPEG file contained compression artifacts in the form of perfect blocks, which is obviously not good.
Here are my settings in detail:
Output: NTSC, 1920x1080, 23.976 fps, Progressive, Quality 75 VBR, 2 Pass, Min 25.00, Target 30.00, Max 35.00 Mbs. Aspect Ratio: Square Pixels. Profile: Main. Level: High. I've checked "Render at Maximum Depth" and "Use Maximum Render Quality". The GOP settings (I don't know what these are) are at the default: M Frames: 3 is checked and N Frames 12 is selected from drop down. I have not checked "Use Frame Blending"
For the H.264 in Adobe Media Encoder, I generated a file that was 20gb. For this one I did an experiment and set the key frame distance to 1 because I think that the grain really makes it difficult for the compression scheme to reuse data productively from frame to frame. My original film is so grainy that literally every part of every frame is different from the last.
My settings were:
Output: 1920x1080 (1.0), 23.976 fps, Progressive VBR, 2 Pass, Target 30.00 Mbps, Max 35.00 Mbps. TV Standard NTSC. Profile High. Level 4.1. I've checked "Render at Maximum Depth" and "Use Maximum Render Quality".
Here's a brief summary of my conclusions and some questions:
My own comparisons involved examining individual frames in each file to compare the compression artifacts. As I mentioned, the MPEG file I made with AME had perfect blocks in certain frames, so this seemed to be the worst of the three.
Overall the H.264 file that I made with AME with the 1 frame key framing seemed slightly better (it's a bigger file, so it's not surprising that it would look a little better, plus the keyframing on every frame was in place on only the AME file). Certain problematic frames were handled better, though they still contained visible artifacts. But then certain other problem areas were actually worse in this larger file. So I can't conclusively say that the AME file is much better than the other.
First of all, do Compressor and AME actually work differently from each other, which would necessitate my continuing to compare their H.264 encodes against each other? It makes sense that the AME file looked slightly better because it's a larger file so I really can't compare them accurately. But if the compression method is the same between the two programs then at least I can rule that out as a variable and choose just one program or the other for further tests.
Should I stick with H.264 encoding or should I give MPEG2 encoding another try?
Does setting the keyframing to every frame make sense or am I actually hindering the program's ability to encode productively and ending up with a worse result?
Are there other tools besides AME and Compressor that I should be considering?
Thank you very much!
I am working on a project that is almost identical to this. I would really like to save weeks of headaches and experimentation, which is usually what it takes to get grainy ProRes film scans to look even halfway decent on Vimeo. I would really really appreciate it if someone who knows how to achieve Criterion-quality professional Blu-ray encodes would chime in here.
Also, Vanessa, since your first post was a few months back, if you have since developed a workflow, I would be very grateful if you would post it.
Thanks so much,
I have not yet been able to perfect my workflow unfortunately. For exhibition I've been using an blu ray encoded with compressor. I did speak with a couple of people who felt that a higher level hardware encoding set up would achieve the best results, specifically using sony blu code (see message below which I received from a lab in NYC - and which introduces new technical challenges to the equation). However, some people think that software encoders like AME and compressor have caught up or surpassed seldom updated hardware encoding set-ups, and obviously paying a lab to master files for you is very expensive.
We will eventually be doing an actual release on blu ray so hopefully someone has some advice to offer!
Thanks for getting back to me. Not sure if you have come across any information regarding Criterion's testing of the open source X264 codec for Blu-ray encoding. Apparently they said it was as good as anything that costs $100,000 and still requires HDV tape--haha what the flip!?!? I mean I am a Luddite who loves physical media but this is 2014 for heaven sakes--(http://www.x264bluray.com/).
I have used the X264 codec extensively in the past via MPEG Streamclip as a front end to incorporate a 2.2 gamma tag into exports for the web to bypass H.264 gamma shifts, however apparently if you really know how to use this thing you can do spectacular things with it for Blu-ray. There are specific tags and algorithms for film grain compression. Unfortunately it kind of seems as though you need to know how to program to really get the most out of it (even with advanced front ends that give you access to all the features such as Hybrid...which I tested out a ton yesterday and could never get to export anything but mpeg2 file for DVD).
HAVING SAID ALL OF THAT...
After a bit of testing of my own with AME CC and Encore CS6, I am very very close (or so it seems) to a workflow that actually knocked my socks off last night upon watching my first test disc...I mean knocked them off...and anyone who knows me would tell you that that is almost impossible to do :)
I MEAN LITERALLY CRITERION BLU-RAY CALIBER.
BUT I need to figure out one issue... http://forums.creativecow.net/readpost/3/954694