The Auto settings on my Topaz enhancements always seem to select the highest bitrates (around 180 Mbps on average), but the lower bitrates produce quite noticeably lower quality outputs. AVCLabs video enhancer AI produces reasonably sized files (only around 12 Mbps on average, which is common for commericial 4K videos), with only marginally lesser quality than the Topaz AI. I’d much rather have a 1.7 GB file than an 18 GB file, especially if the output quality is still very high.
How can I replicate this kind of video rendering optimization with Topaz?
Personally, I do a trial export of about 5 to 10 minutes of video with the output set to 180Mbps. Then I look at the actual bitrate generated. Depending on the source, the model used and the amount of noise that survives, I find most of my 4K upscales come in at between 16Mbps and 32Mbps.
Not really a solution, but a workaround:
Since Topaz’ video encoder isn’t really optimized and only sparsely configurable I normally choose a HQ big file-size output that should be transparent in quality (e.g. h265 120Mbps).
After the Topaz export has finished I then downsize the exported file with handbrake that allows you to get real high quality in a moderate file size (e.g. using h265 RF 24 with preset medium or slow).
Since handbrake runs on the CPU cores (or the special HW encoder if you chose hardware encoding) and Topaz runs on the GPU and/or Neural engine you can run both tasks simultaneously without much speed loss for either.
Please disregard this thread. I’ve found that manually selecting the lower bitrates still produces a high-quality output, but with a reasonable size. They don’t look ‘awful’. Thanks again for your great product!