Uncompressed 8-bit 4:2:2? or Uncompressed 10-bit 4:4:4?
I’m in the process of capturing the full resolution HD footage for my project’s online sfx and final output work. This is a feature film project that will end up on film for theatrical release and also on DVD, TV, etc. for ancillary markets. The system used is as follows:
- Camera: Sony HDW F-900 HDCAM CineAlta
- Tapes: Sony BCT-40HD
- Shooting format: 1920x1080 25p
- Workstation: Mac Pro 1 (Dual-Core Intel Xeon 3GHz processors –16GB RAM – NVIDIA Quadro FX4500 graphics card – MAC OS X Tiger10.4.10 operating system)
- RAID Storage: G-TECH GSPEEDXL12 Dual Host 4Gb FC with ATTO Celerity FC-42ES host adapter card
- External monitor: JVC DT-V24L1D broadcast LCD monitor
- Software: Fincal Cut Studio 2, Shake 4, Studio Artist 3, Adobe Creative Suite CS3 Production Premium.
- Capture card: Blackmagic Decklink HD Pro PCIe
- Capture VTR: Sony HDW-M2000
Blackmagic Deck Control utility software was used to capture the above footage in the above system, using the following settings:
Easy Setup settings:
- Blackmagic HDTV 1080i50 -8bit (there is no Easy Setup for 1080p 25Hz)
Serial Device settings:
- Serial Port: RS422 Deck Control
- Frame Rate: 25
- Video Input: Blackmagic HDTV 1080
- Source: Blackmagic HDTV 1080i 25Hz
- Compression Type: Uncompressed 8-bit 4:2:2
- Motion Frames Per Second: Best
I have a question regarding all the above:
I’m not sure what my compression type should be for the above footage. Uncompressed 8-bit 4:2:2, Uncompressed 10-bit 4:2:2, Uncompressed 8-bit 4:4:4, or Uncompressed 10-bit 4:4:4?
The datasheet for the Sony HDW F-900 HDCAM CineAlta camera say the following under the VTR specs section:
“Quantisation: 12 bit/sample of input-output signals (8 bit sample for internal compression process)”.
Wikipedia say the following about Sony HDCAM tapes:
“HDCAM, introduced in 1997, is an HD version of Digital Betacam, using an 8-bit DCT compressed 3:1:1 recording, in 1080i-compatible downsampled resolution of 1440×1080, and adding 24p and 23.976 PsF modes to later models. The HDCAM codec uses non-square pixels and as such the recorded 1440×1080 content is upsampled to 1920×1080 on playback. The recorded video bitrate is 144 Mbit/s”.
I’m not sure if the above camera and tape specs are relevant to my question about the compression type for the post-production capture settings, but I thought I would post the info anyway in case if it shed some light on the subject.
I guess what I’m trying to ask is this: does it matter if the camera and the tape captured an 8-bit 4:2:2 image during the shoot and I decide to capture this image in post-production as 10-bit 4:4:4? If it does indeed matter, then, given the above camera and the tape that I used in the shoot, what should my decision be on the compression type when I’m capturing my footage in post production?
[Ayman Farahat] "I guess what I’m trying to ask is this: does it matter if the camera and the tape captured an 8-bit 4:2:2 image during the shoot and I decide to capture this image in post-production as 10-bit 4:4:4? If it does indeed matter, then, given the above camera and the tape that I used in the shoot, what should my decision be on the compression type when I’m capturing my footage in post production?
If you captured 8bit 4:2:2 then that's what you should ingest.
Capturing 4:4:4 will not really be of any benefit since the information simply is not there to begin with. Could be a little help if you are going to color correction, but then you have the issue of outputting your project at 4:4:4.
good luck with your project.
Many thanks, Mark, for your response. There will be a significant amount of compositing, fx, coloring, etc. (the whole film was shot on green screen) and this is why I'm thinking of 10-bit vs 8-bit and 4:4:4 vs 4:2:2 although, as you rightly say, the information might not be there in the original footage. The thing is... I'm not sure about the information in the original footage, probably because of my technical naivety, and I don't know whether the camera and the tape used (as I described them in the first message) captured an 8-bit or a 10-bit image at 4:2:2 or 4:4:4. So... I'm thinking... there is nothing to loose if I go for a 10-bit 4:4:4 capture in post-production... if the information is not there, then I have only lost some disk space. This is, of course, unless there is some technical reason/difficulty that I'm missing, again because of my technical naivety, which would "prevent" me from capturing an 8-bit 4:2:2 footage as 10-bit 4:4:4... any thought on that?
[Ayman Farahat] "So... I'm thinking... there is nothing to loose if I go for a 10-bit 4:4:4 capture in post-production... if the information is not there, then I have only lost some disk space. This is, of course, unless there is some technical reason/difficulty that I'm missing, again because of my technical naivety, which would "prevent" me from capturing an 8-bit 4:2:2 footage as 10-bit 4:4:4... any thought on that? "
Well... first off... do you have a 4:4:4 monitor to check your work?
Also, can your editing/composite system/software handle the 4:4:4?
My suggestion would be to ingest a short test at 4:4:4 and run it through your ENTIRE workflow.... all the way back to final deliverable.
This way you will find any potentially project-killing roadblocks.
We recently colored a feature in 4:4:4 and it was quite a learning experience.
Good luck with your project.
Good luck with your project
thanks, Mark, you make some good points there... cheers! What about 10-bit 4:2:2 as opposed to 8-bit 4:2:2? any thoughts on that? From your experience, is it reasonable to expect from 10-bit 4:2:2 a better gain (than with 8-bit 4:2:2) in terms of image quality and perhaps less trouble (than with 10-bit 4:4:4) in terms of workflow difficulties?
...not sure what a "4:4:4 monitor" is...there are analog component, SDI, DVI and HDMI monitors out there...all of them handle 4:4:4 as they have to to create a color picture. Whether or not the footage is 4:4:4 or 4:2:2, once it's played back it has to account for all color channels at the monitoring device...
The key here is this...nothing prevents you from capturing 10 bit precision 4:4:4 from your HDSDI connection coming from the HDcam deck...inside the HDSDI cable, the video is considered "digital baseband" which is a mutated analog term meant to say that the SDI pipe has no knowledge of compression or subsampling...it carries an image format.
However, what is coming from your deck, even after it's "decompressed" from the very aggressively compressed HDcam format, is only 4:2:2...I think Sony contends that it's 10 bit but that's a stretch as only 8 bits were saved...
If you decided to capture 10 bit, you will probably gain something in the compositing process as 8 bit video has 256 discrete steps within each color channel, while 10 bit has 1,024. You won't create information that wasn't there to begin with (I need to stress that), but you may get some smoother edges as the footage is re-processed from the incoming SDI signal.
...but stay 4:2:2...you will already have a larger file with 10 bit, I really don't think you'll gain anything by moving to 4:4:4.
Thanks very much, Tim, your answer addresses and confirms my thoughts on this... Sony left me confused between the 10-bit and the 8-bit quality of my footage... my thought was that I probably have got 8-bit footage with their CineAlta F-900 camera, but ingesting that footage in 10-bit in post production will probably give me smoother edges and maybe better compositing... glad to hear experienced thoughts on this point that echo mine.
This forum is awesome and you guys are brilliant in your experience and support. This is the first time I use this place and I'm really grateful for your thoughts and help!!
I had another question on the same project, which I have posted under a different section (Blackmagic Decklink)... I'm using Decklink HD Pro PCIe card to capture this footage, but to my surprise (and frustration) I've discovered that Blackmagic people deliberately do not provide support for capturing 1080p/25 (their card supports only 1080i/50); they claim that 1080p/25 is an extremely rare format... and guess what... my footage is 1080p/25 and the last thing I want is to convert my progressive footage into an interlaced one during the capture process (I understand that sfx work yields better/more reliable results when it's done on progressive rather than interlaced footage). I'm about to take the plunge and spend a couple of grands on the Kona 3 card because it supports the 1080p/25 format, but I was hoping that there is someone out there who might know a way of using the Blackmagic Decklink HD Pro card to capture 1080p/25 footage without converting it into 1080i/50 footage... any thoughts on that?
Many thanks again for all your help!
[Ayman Farahat] "Blackmagic people deliberately do not provide support for capturing 1080p/25 (their card supports only 1080i/50); they claim that 1080p/25 is an extremely rare format... and guess what... my footage is 1080p/25"
Yeah...funny thing about that. The F900 doesn't actually record "progressive" per se.
The F900 (among others) records what's usually denoted as "PsF" or "Progressive Segmented Frame" or one more time, a "Progressive frame recorded and stored as interlaced fields."
The basic idea here is that even though you shot the footage as "progressive", it's stored as interlace. Your editing system can read it as 50i on ingest because technically it is.
The issue comes in when you're trying to post this stuff...what will you output to? If you intend to go back to HDcam, posting in 50i might not create an artifacting problem...but this becomes the potential issue. ProRes has several 25p 1080 setttings, but I've not worked with the BM cards enough to know if they'll even hand the footage off to ProRes...not to mention that ProRes is another form of relatively aggressive (albeit decent quality) compression...
Premiere Pro can be custom set for 25p/1080, but you'd probably have to capture the footage first, then start a new project and import it as that custom setting won't communicate with the BM card if there is no setting available for it. Also, if you are editing in a "custom" mode that isn't supported by your card, you can't use it to monitor your timeline either...
While I know it's painful to spend money (particularly to replace items you already have), I use AJA products myself and I run into situations constantly where the ability to handle huge variations in frames sizes/rates AND the ability to convert on the fly (edit HD, monitor SD....edit 1080, monitor 720p...on and on...) solves a problem for me that is a roadblock for someone else...
My 2 cents...
Good luck with it.