PDA

View Full Version : CineForm V3.2.3 released



David Newman
01-05-2008, 09:56 AM
New builds for all CineForm packages are now available. As the 4K products are still in beta, we bumped the trial licenses another 15-days. All should be official out within the next two weeks.

galexander
01-05-2008, 10:24 AM
i was reading about the work flow, why isn't there a 4k project setting? maybe i don't want to scale my image or over sample. if i wanted to work with a 2k camera, i'd look at the silicon image solution. 2k is only ~16mm film, why degrade the data?

why not instead work at 4k and support outputs like mp4/x264?

also will this work with premiere cs2?

do you have an SDK for WinDoZe? i'd like to hook in and write out a different file type than QTs or have to always wrap to mov. i like to use avisynth and virtual dub more than commercial codz.

JustMe
01-05-2008, 10:50 AM
i was reading about the work flow, why isn't there a 4k project setting? maybe i don't want to scale my image or over sample. if i wanted to work with a 2k camera, i'd look at the silicon image solution. 2k is only ~16mm film, why degrade the data?



You can just make a 4k setting, if you have the cpu/mem to do it.
If you did look at the SI2k you would be looking at CineForm..

David Newman
01-05-2008, 11:03 AM
galexander,

Clearly I our workflow needs to be explained better as you have misunderstood the nature of oversampling. A 2K timeline does not impact your DOF (16mm vs 35mm) nor in fact your ability to edit 4K. We have not included a 4K preset for these reasons:
1) It only takes 30 seconds to create your own.
2) 99% of users will not need to edit in 4K, even with 4K sources.
3) If we had a 4K preset 95% users might mistakenly use it when 2K timeline would be better for them. There is no degrading of the data.
4) This is editorial. We currently recommend the 4K masters be made in After Effects. We are working to enable 4K 32-bit float masters out of Premiere (that is why it is still beta.)

Yes Prospect works with either CS2 or CS3.

Elizabeth Lowrey
01-08-2008, 09:46 AM
I've registered twice to obtain the trial version of NEO 4K. I didn't receive the promised email with download link either time. My email account is active and receiving other mail without problem. If someone at Cineform could help me out, I would appreciate it.

David Newman
01-08-2008, 10:00 AM
Please check your spam folder. I just confirmed that NEO 4K trial email is working. However I will PM you the direct link, just in case.

galexander
01-12-2008, 07:24 PM
New builds for all CineForm packages are now available.

question for you thanks.

for money no object, cash in hand, something that works today, right here, right now.

what would you suggest for a complete 16-bit work flow? maintaining 16-bits throughout the entire process.

what would you suggest for a complete 10-bit work flow? maintaining 10-bits throughout the entire process.

thanks :)

David Newman
01-12-2008, 08:35 PM
Really 10-bit and 16-bit are the same as 10-bit data is often processed at 16-bit as it convenient in Intel procs. The chooses are 8-bit (say no), 16-bit integer and 32-bit float. When developmenting a RAW image you are likely to make white greater than white, and you don't want to clip this until your look is set, so 32-bit float is the way to go. 4K at 32-bit float RGBA, is a little painful at 128MB per frame, so be warned.

As for money no object, that is not a CineForm question, try Discrete. :)

galexander
01-12-2008, 10:14 PM
Really 10-bit and 16-bit are the same as 10-bit data is often processed at 16-bit as it convenient in Intel procs. The chooses are 8-bit (say no), 16-bit integer and 32-bit float. When developmenting a RAW image you are likely to make white greater than white, and you don't want to clip this until your look is set, so 32-bit float is the way to go. 4K at 32-bit float RGBA, is a little painful at 128MB per frame, so be warned.

As for money no object, that is not a CineForm question, try Discrete. :)

<html>
<body bgcolor="#0f0adf">
<font size="3" face="Courier" color='white'>
<p>
wow, thanks for the quick response.

i think i'm re-evaluating whether 4k workflow is viable these days.

IMHO QT's libs are flakey and not implemented and well, in layman's terms, QT sucks donkey balls for anything greater than 8-bit, the 16-bit or greater depth appears very, very poorly implemented. i'm not talking only theory but current reality.

http://www.reduser.net/forum/showthread.php?t=7283

the 1080p should be at "least" 10-bit but has been chopped to 8-bit. the frame error 8961 is mostly likely due to non-square pixel/frame, but why 8-bit?

21:18:23.30 ERROR: QT Codec: Failed to compress frame (error -8961)..
: Frame: 1920x1080 - 8 bit
^^^^ wtf???

http://www.reduser.net/forum/showthread.php?p=135663


while these are not Cineform issues per say and i'm not asking you to fix any of QT's problems. as a whole, if i'm going spend this kind of $$, i want some confidence that i'm getting something usable. i might as well just keep using IDL and Matlab.....


how easy is it to integrate Cineform into another independent workflow? say i have true 16-bit MONO (no deBayer or demosaic) at 1920x1080, 30fps, GiGe? could you do hook into the GiGe drivers in real-time?


i see tons of posts but want your CTO guru opinion specifically about the difference between having a 1920x1080 16-bit data versus the bigger but numerically inferior 4k @10-bit? smaller but much more information versus huge without as much. so if i work with smaller better then scale up to 4k or film, let's say i can hack a pretty good and fast interpolation routine.


1920x1080@16-bit integer ~ converted to 32bit float ~32MB/frame??


thanks!!

</font>
</body>
</html>

David Newman
01-13-2008, 09:38 AM
Yes QT can be flakey. RedCine should be using the pixel format 'b64a' which 16-bit per channel when accessing the CineForm codec. I have the feel that the 32-bit XP application is running out of memory and doing 8-bit I/O on some systems (total guess at the moment.) We will be investigating, but unfortuately is working fine on our test systems (so it is system dependent.)


how easy is it to integrate Cineform into another independent workflow? say i have true 16-bit MONO (no deBayer or demosaic) at 1920x1080, 30fps, GiGe? could you do hook into the GiGe drivers in real-time?

This gets into areas way not Red related. Email me dnewman at CineForm.

Could you please restate the 16-bit question, as I'm not sure what you are trying to achieve.

galexander
01-13-2008, 10:05 AM
Yes QT can be flakey. RedCine should be using the pixel format 'b64a' which 16-bit per channel when accessing the CineForm codec. I have the feel that the 32-bit XP application is running out of memory and doing 8-bit I/O on some systems (total guess at the moment.) We will be investigating, but unfortuately is working fine on our test systems (so it is system dependent.)

This gets into areas way not Red related. Email me dnewman at CineForm.

Could you please restate the 16-bit question, as I'm not sure what you are trying to achieve.

hi thanks!

sent you a private email via hotmail, so it might get dumped into your junk mailbox.

i put up a new thread with a poll to hopefully find out exactly who has a complete 10-bit pipeline working and am now very interested if you specify what your system is hardware/software, codecs revs please?

cheers!

Mike McCarthy
01-13-2008, 01:28 PM
This is a bit of a random question, but when wandering through the RedCine forum, I found an update on the 4K 2:1 vs. 4K 16x9 status. RedCine supports both. My Cineform tests were done with the 4K 2:1 demo files. I see 4K 16x9 as the ideal way to shoot. Is 4K 16x9 fully functional with Prospect4K? Are there any known limitations with either setting, relative to the other?