So I used Video Enhance AI to upscale a 1080p video to 4K. Due to the length of the video and the past version crashing. I went with exporting to .tiff with the plan of taking all the frames and encoring them into a mp4.
Hasn’t worked out for me so far. I have the majority of the frames on one hard drive, the rest on another once I used up the space of the 1st drive.
Download FFmpeg and have been trying to create the MP4 without any luck.
I made a concat file called frames.txt that lists all the frames to be used.
I then run this command:
ffmpeg -r 29.97 -f concat -safe 0 -i frames.txt -vcodec libx265 -crf 20 -pix_fmt yuv420p F:\output.mp4
and I get this for output
[concat @ 000001a5638ec840] Impossible to open 'J:\video\000000.tiff'
frames.txt: Protocol not found
Did you mean file:frames.txt?
I have no clue to why this is happening and I’m new to FFmpeg. But it clearly opens the frames.txt file and it starts to read the 1st frame. The path is correct and so is the file name.
I’m not similar with the frames.txt file, but I use something like this:
ffmpeg -r 29.97 -f image2 -i %06d.tiff -i audio.aac “output.mp4”
-r = frame rate
-f image2 = tells ffmpeg that the format is an sequence of images (ie. 1.tiff, 2.tiff, 3.tiff)
-i %06d.tiff = describes the sequence format (a six digit number with leading zeros followed by the file extension as .tiff)
-i audio.aac = tells ffmeg I also want to use this audio file as input
Add other parameters to tweak the output encoding type etc
I’m not sure if it’s possible to use a frames.txt as input.
If you really have issues hard drive space, you can rename the second portion of files from 010000.tiff etc back to the original sequence starting at 0000000.tiff, then do a ffmeg concat to join 2 or more video files back into one.
I haven’t tried making tiff’s into video yet but was planning on giving virtualdub a try just my 2 cents
Use VirtualDub2, it’s a lot more easy. Drag and drop the vey first image and that’s all, all the other images are normally automatically loaded. Then set the framerate of your video if different from the default settings, import the sound from the original video. And finally choose the codec (video + audio) you want to use for your final output.
The only issue is that image sequence process is not multithreaded and can be slow (if you are encoding HD or 4K with x265 this will not be a problem because the encoding process is slower than the loading of the images. Using Avisynth+ and the ImageSource internal filter is twice faster.
Premiere Pro is a beast to render image sequences. If they are stored on a SSD then it’s even faster. But PPro can not import 16 bit TIFF images. Only 8 bits TIFF or PNG.
Premiere Pro should be able to import Tiff 16 bit files. If you take one of the Tiff 16 bit files from AI, open it in Photoshop, and immediately re-save it as a same Tiff 16 bit file, you can then import it into Premiere Pro with no problems.
There is something wrong with the way AI writes the Tiff 16 bit files.
