Blackmagic Forum

Have any of you DR 17.b6 testers used any of the deinterlacing tools/options?

I’m on a doc (DR 16.2.7) which has multiple stock footage clips (from the year 1920) with conversions from original 16-26 fps to 59.94 and then 29.97df.
Wondering if DR 17 can at least deinterlace as well as Nucoda or others. I’m aware the variable hand cranked footage won’t respond well to 3:2 pulldown removal. The deinterlacing will be the minimum necessary to pass QC. I’m reluctant to upgrade to 17 until stabilizes around verso 17.2.
Anyone have a positive experience with that “newer” deinterlacing option?


Sent from my iPhone using Tapatalk

Dario, if you want the highest possible software deinterlacing, have a look at QTGMC from Avisynth, which is a free scripting system for video.
I am using the tool Staxrip which has GUI frontend for Avisynth scripts and brings all the tools necessary. Thanks to FFMPEG you can even export into a ProRes mov on your PC.
https://www.videohelp.com/software/StaxRip

It takes a bit time to understand how things work, but the results are well worth it.

I have done some deinterlacing tests with R17b6, it's ok but can't touch QTGMC quality.

If you don't want to dive into Staxrip/Aviynth or don't have the time now, you can PM me a download link of some sample clips you need to process and I will put it through my pipeline so you can see if it will do what you need.

It all depends on how the footage was treated and edited before. If it was a straight transfer at a fixed speed you may be lucky to find a pulldown pattern that fits. I'm not sure DR 17 can do this, we used to do such stuff with AEFX.

If it's completely irregular, things can get quite complicated with shot-by-shot corrections needed. Your best bet might be then to apply a good deinterlacer to the whole footage without trying to reconstruct original frames (assuming that at least the field dominance is regular). As Robert explained, QTGMC is simply the best, but the new neural filter in DR 17 is doing a decent job, far better than the one in 16.

The footage is from a 1920 film documenting an event. Testing is showing that the footage has hand cranked varying frame rates and has been converted at least twice into different rates (info in my original post) I’ll be trying the suggestions for DR 16 listed above as well as hopefully testing out the DVO tools.
I’m going to avoid 17 for now. I only have one main machine to run DR on while working remotely.

I thank everyone for taking the time to make great suggestions.

Here’s the link to the LGG thread which had good comprehensive answers as well. Any continued discussion should go there. For future reference for others.

https://liftgammagain.com/forum/index.p ... ing.15361/


Dario Bigi, CSI
HP z840 - Dual 14 Core 2.6GHz - 64GB RAM - RTX 2x2080Ti - Internal SSD 8TB Raid. Windows 10 Pro 1903 - Resolve & Fusion Studio - FSI CM250 - Panasonic FZ1000 - Mini Panel

I've just started a new post "Enabling Neural Engine Deinterlacing in Resolve 17". which may be of interest.
I'm now wondering whether I should have posted it here rather than starting a new post. I'll leave it to the moderators to decide.

Hi Stan,
In my original post I stated I was looking to avoid going to v17 until it stabilized and is out of beta. Upgrading being a possible consideration. Other colorist offered viable alternatives using v16 in the LGG thread. I posted the link for others to follow if they are as cautious as I.


Dario Bigi, CSI
HP z840 - Dual 14 Core 2.6GHz - 64GB RAM - RTX 2x2080Ti - Internal SSD 8TB Raid. Windows 10 Pro 1903 - Resolve & Fusion Studio - FSI CM250 - Panasonic FZ1000 - Mini Panel