I had some old DV .avi 29.97fps NTSC interlaced footage, which I fed through Hybrid to deinterlace it.
Hybrid bobbed it up to 59.94, and it’s now 640x480 square pixel, ready to feed into Starlight Mini (7.1.5) for a 4x upscale.
My question is, can I convert this back to 29.97 before I feed it back into Starlight Mini. There appears to be no option in SLM to convert it back to 29.97, - it maintains the 59.94, which means it takes a lot longer to render.
Or, do I risk having motion issues, and losing some detail?
Your best bet would be to halve the framerate while processing it with Hybrid. Maybe the SelectEvery filter (assuming that’s possible in Hybrid), or there might be an option when de-interlacing to not double the framerate.
The Bob flag into hybrid does this. I do doubling first and then I check result into Virtualdub picking some frames. When there is no movement into the additional gained frames, or if they look worser than the others, then I don’t do “Bob” doubling. It’s really easy to see, just look at a few frames in series into Virtualdub, then decide. When you do double, then don’t half it for Topaz, this is a loss, so yes you have the double upscale time when you double it
In some cases, doubling brings enormous gains in movement, in other cases it brings nothing or can even be a deterioration. It really varies from case to case, so I recommend looking at the result frame by frame beforehand.
I’m not familiar with Virtualdub, so might lean towards leaving it at 59.94 and taking the pain of extra upscale render time. I might look at the 59.94 footage before I upscale in case there’s any deterioration, but I think bobbing usually improves things.
I just downloaded version 7.1.5, the discouraging thing is that I am only getting 0.2fps on render.
I moved the # of processes up to 2, and I have a new 5090 rig. I demoed 7.1.0 and I was getting 0.3 or 0.4fps.
Any comments on this - did Topaz change things resulting in a slow down, or are there any tweaks I can make to get quicker renders?
You don’t have to do anything into Virtualdub just drag deinterlaced result into it and watch, use arrows keys going from one frame to the other and then you can see what deinterlacer did (with doubled frames) so it’s really easier than it sounds
I know most say doubling is better, I say no it pends on! When I double frames of Star Trek Voyager seasons then I get additional doubled frames that are worser copies of the original frames, without any “movement gain” so static worse copies of exisitng frames and this brings me nothing.
But it could also be you gain additional frames that looks good and have movements in it, then using Bob flag while deinterlacing is king.
With your Virtualdub workflow which you describe above, are you saying you end up with portions of the video that are motion intensive in 59.94, and other portions of the video which are more static in 29.97?
If so, how do you knit together all the video if some of it is in 29.97 and other parts in 59.94?
Hi, no I don’t mix 29.97 with 59.97. I do the full video 29.97 OR doubling 59.97 (of course there exist content that is progressive and interlaced mixed, but this are special cases)
But how do you know 59.97 is better or not? The answer is doing a visual check of deinterlaced doubled result. I always do doubling first, but I’m not sure keeping this, until I have visual checked some frames, the new doubled frames deinterlacer has produced.
So you can import doubled deinterlaced video into VirtualDub just for a visual check use it as video viewer “how good looks the new frames?” Based on this I decide keep double framerate or not, when not then I deinterlace again into Hybrid without the “Bob” flag.
In Motion there is an Optical Flow frame blending that is supposed to be good with motion heavy scenes. Have you come across FFmepg minterpolate? Thats another option I might look at.
I’ve looked at 60fps footage, compared to 24fps, and 24 actually it generally looks more natural to my eyes.
Hi, I don’t know this applications, but all interpolation tools have the problem of generating blury frames when the footage contains already unsharp frames. This can lead to unnatural look, “fluid” movements into video. TVs can do also live-interpolations. When the backround is unsharp the new generated interpolates frame of it gets more blury, while the sharp foreground does not, well known as the bad “soap effect”
But Bob deinterlace is different, you really can gain additional good frames, but sometimes does not and just blows up file size, pends on the source.
