I believe that trim by frame specification can be supported.
FFMPEG can also be trimmed by specifying the frame.
-vf trim=start_frame=10:end_frame=30
There may be some implementation issues such as how to handle frames on the GUI side, how to work with vfr, and how to handle audio (audio is not in terms of frames, but samples).
I just use losslesscut to trim anyway. Would be nice to see the same functionality in Video AI though!
Thanks, yes, it’s exactly that !
I wrote a CineForm for encoder.json.
{
"text": "GoPro CineForm HD",
"encoder": "-c:v cfhd -pix_fmt yuv422p10le",
"ext": [
"avi"
],
"transcode": "aac -b:a 192k -ac 2",
"os": "windows",
"minSize": [1,1],
"maxSize": [16255,16255],
"maxBitDepth": 10,
"doNotScaleFullColorRange": "transcode"
},
For reference, here are the FFMPEG option settings.
"C:\Program Files\Topaz Labs LLC\Topaz Video AI\ffmpeg.exe" -h encoder=cfhd
Encoder cfhd [GoPro CineForm HD]:
General capabilities: dr1 threads
Threading capabilities: frame
Supported pixel formats: yuv422p10le gbrp12le gbrap12le
cfhd AVOptions:
-quality <int> E..V....... set quality (from 0 to 12) (default film3+)
film3+ 0 E..V.......
film3 1 E..V.......
film2+ 2 E..V.......
film2 3 E..V.......
film1.5 4 E..V.......
film1+ 5 E..V.......
film1 6 E..V.......
high+ 7 E..V.......
high 8 E..V.......
medium+ 9 E..V.......
medium 10 E..V.......
low+ 11 E..V.......
low 12 E..V.......
2020 iMac,128 GB Ram, 10th gen 10 Core i9, 5700XT 16 vram, source dv to progressive, enhance and upscale from 640 to 1920 with both Artemis low quality and and Proteus manual. Source on very fast SSD and Exports on Very fast SSD. I’m am seeing amazing performance and quality improvements. I’ve seen 8 frames per second exports when previously I might get .something other output. That’s incredible. Also, I’ve been working on a creative directors spots and the problems with faces for example has dramatically improved. He may be able to show at 1920 progressive what was once 640 interlaced to clients now given the Improvement you folks have made. Well Done! And godspeed to further quality improvements. Beta v3.1.1.1.B
Prores is lossless so is are TIFF files.
tbh all i want to see is deinterlacing as a standalone feature so all models can benefit from it. instead of it being its own model.
Thanks, that’s exactly what I needed!
I also added selections for medium and high quality as well (for FHD and UHD, respectively), since the default is the highest quality setting if none where specified.
I had actually bought a Cineform license a few months before GoPro bought them out years ago and back then had talked to the guy who had invented the codec about the quality levels to use for various resolutions.
His advice had been that with the way the wavelet encoding worked, while using higher quality settings resulted in a larger file, there was no actual gain in quality for HD material above the medium setting, and no gain for 4K (UHD) material above the high setting. And while 8K resolution was a pipe dream at that time, that would be where you would use the film setting of the codec.
Over the years that’s shown itself to be true in my experience.
Again, thanks!!
James
Amazing speed difference. What’s your setup? I see a jump in mine, depending on the models I use, but for me I’m using Mac Pro M1.
I have a last gen beast: RTX 3090 FE/Ryzen 3900x/64 GB DD4 win 10 64 Pro, it should last me another few years though!
I’m new to this and honestly would have to look it up! It sounds powerful, whereas mine is maybe entry level as far as AI upscale is concerned.
Yeah, it took me an ignorant amount of time saving up for it, so I’m in no hurry to upgrade, hell, it would take me a year or so of saving up to upgrade my rig to the current gen!
It’s not actually lossless. Wikipedia says it’s a high quality visually lossless codec.
So not totally lossless, it just looks the same visually (but may be losing details that you can’t easily see but which might be useful later in editing). eg. a totally lossless codec could be a bit better for keying.
I know…but I would prefer the UT Video option to be possible
No codec is “ lossless” per se. Anytime VEAI exports it is re-encoding no matter to what or the codec. Anytime you re-encode you’re loosing a generation. Therefore by definition the export is not lossless. No export is. Practically speaking though exporting to Prores HQ and TIFF 16 bit is lossless. In other words and as you mentioned, one can’t tell the difference between the original and the export-even on a big screen. Even exporting to 4444 isn’t lossless. Yet, practically speaking if you export from the raw to 4444 the export will still be perfectly fine for both VFX work and color grading. 422 HQ is great for keying although 4444 or the raw is ideal. Though never as good as the original source, the difference is so minuscule, practically speaking, that it’s irrelevant. You make good points however. I’m just being pedantic.
You can use something like handbrake or ff-Works as standalone to do that.
100% lossless doesn’t exist…but UT Video I think it’s the best in this…TIFF doesn’t work for me because of how I edit afterwards. I think that UT Video is not due to copyright issues… but since it is (ffmpeg) I thought that it could be implemented in some way.
Fair enough sir👍
Thanks so much for all the improvements.
Are there any plans for HDR functions?
After this update. I’ve been having trouble with the software, it’s not exporting videos. and some get the message “Unknown error” - “last FFmpeg messages”. I’m sure if I’m the only one having this problem. But, I’d like some feedback on this.
