Blackmagic Forum

Greetings!
This is my first post here. Trying to establish a workflow for getting DV 720x480 29.97 into a 720HD timeline with optimum results. Using 17.2.1 Studio.
A little unclear on a few things, like Super Scaling. Does this work on SD footage or does the standard scaling take over in the event the footage is not HD?
Should I choose to enable interlace processing? If I don't does this mean I can use the Neural engine setting(if selected) and it will create whole progressive frames? Tried quite a few settings and tons of test renders and am still getting some artifacting on the export. Tried de-flickering didn't seem to help.
Footage is pretty simple and I have applied some noise reduction and basic CC and mild sharpening. Most of the noise is in the luma channel.
I know many of you have been through this, but I would be grateful for any workflow tips you may have to make sure my timeline and project settings are correct.
Thanks in advance!

SuperScale works for SD too.
Neural de-interlacing is decent and since 720 is always progressive, I'd set the timeline to do progressive with twice the fps to preserve the temporal resolution.

If you are looking for something even better at the cost of long processing times, try Topaz Labs Video Enhancer AI (Dione model).

That Topaz thing is very impressive Uli, do you need Nvidia RTX gpu hardware to make proper use of it?

It runs on my Macs with only medium-class AMD GPUs too. A RTX would be faster, though.
In my experience VEAI is great with natural textures. Human faces can be problematic.

Thanks Uli. Pricey but I may give it a go - is it that much better than DVR's neural effort?

Thanks for the heads up. I'll try the 59.94 timeline. That makes sense.
Wow Topaz looks pretty incredible. Going to have to see if I can demo that.
Thanks Uli!

Since all of these methods depend so much on content, you'll need to try with your own.

Topaz engine is very sensitive with combination of different source nature/chosen model. Too sensitive. Speed is also a problem if you have many hours of content. On HD sources it’s crazy slow.

Both correct. But since the OP needs de-interlacing too, it's only the Dione model one should try.
VEAI is only worth it if you have relatively short content of high value, personal or general. Otherwise I'd use the features of DR Studio.

So I downloaded the Topaz demo. There were mixed results. The best setting I could come up with was Dione Interlaced Robust v3 for my needs. In the process of trying a batch process now. I have had some quirky things with it such as "can't write file" and folder permissions I am trying to work through. Sometimes when it fails, I can close and re-launch the program and it works fine. I think I have exhausted every setting within DaVinci to get this properly deinterlaced and ready for edit. I would really rather deinterlace and then bring the files back into DaVinci to use clip by clip levels of noise reduction and sharpening along with CC. Tricky stuff. I wish it didn't have to be.
Some of the best deinterlacing I have seen so far was AVI Synth and QTGMC. A bit of a learning curve, but I may explore it. This is a real time killer so far. I'll keep trying. Lots of these methods still have top field like jitter at the top that may be misinterpretation of bottom field first. I guess I will just have to crop the top line out in DaVinci . I appreciate the input as I learn about this!

Topaz VEAI has been updated to version 2.3.0 with a lot of bug fixes and easier comparison of different models, which might help. It does fps conversion too now, might be worth a try. I have not tested that yet.

I have no affiliation with Topaz other than participating in early beta testing under MacOS, which gave me a discount on the product.

Ed Stephens wrote:So I downloaded the Topaz demo. There were mixed results. The best setting I could come up with was Dione Interlaced Robust v3 for my needs. In the process of trying a batch process now. I have had some quirky things with it such as "can't write file" and folder permissions I am trying to work through. Sometimes when it fails, I can close and re-launch the program and it works fine. I think I have exhausted every setting within DaVinci to get this properly deinterlaced and ready for edit. I would really rather deinterlace and then bring the files back into DaVinci to use clip by clip levels of noise reduction and sharpening along with CC. Tricky stuff. I wish it didn't have to be.
Some of the best deinterlacing I have seen so far was AVI Synth and QTGMC. A bit of a learning curve, but I may explore it. This is a real time killer so far. I'll keep trying. Lots of these methods still have top field like jitter at the top that may be misinterpretation of bottom field first. I guess I will just have to crop the top line out in DaVinci . I appreciate the input as I learn about this!

QTGMC gives you full control over field order (top field by default). You need to know it and set correctly as needed.
Cropping lines is a bad way.