I’m attempting to de-interlace and upscale to HD some 1440:1080 60i HDV footage.
I know HDV has rectangular pixels. I see an option in Topaz that allows us to select either “original pixel type” or “square pixel”. I’m afraid I’ve been unable to find any information telling me when or why I might choose “square” or “original”.
I have been making some attempts to upscale this footage and each time I have the resulting image appears as a smaller 16:9 image within the larger 16:9 frame. That is, my image is, you could say, both pillar and letter boxed. Has my not selecting the “square pixel” setting been responsible for this?
So my question comes down to this: when upscaling HDV footage to 1920:1080 HD do I select “square” or “original” pixel?
it will be if the original is 4:3 (assuming you picked Full Frame and not Crop to fill), but the pillarbox would be slimer and the picture area would be wider/larger.
I personally would go for 1440x1080 or 1920x1440 if you have 2k Display (or above) for playback, in case the video is 4:3 display Aspect ratio. as it is more natural resolution for 4:3 then 1920x1080 which is meant for 16:9 Display Aspect Ratio. Unless you going with Crop to Fill option, then 1920x1080 would make more sense.
I’m attaching a screen grab of the footage as it appears in the Resolve windows and its Media page.
The original HDV footage is 16:9 and 1440X1080. When I’ve run it through Topaz I’ve set at “original pixal type”. I’m viewing the footage on a 5K monitor.
If I understand you correctly, you suggest for the media I’ve described 1920X1080, “original pixal type”.
I plan to do some more exoerimenting. The media file I’ve been describing is 6.5 GB, so processing through Topaz has take 10-12 hours each time. This morning I found some HDV footage about 300 MB in length. I’ll be able to run it through Topaz, at various test settings, much more quickly.
To make sure I’m clear. The letter/pillar boxed image in the mesage before this is the HDV interlaced footage after processing through Topaz at de-interlace 1920X1080.
no it’s not, but that resolution is mainly for 4:3 equivalent to 1080p
I was not aware your video was already at that resolution, for some reason I assumed it was SD resolution at 4:3. in that case 1920x1080 sounds right. or you might consider 4k upscale as your source is at a peaty much good quality.
Now that I am thinking about it. since your video is at a high quality, I would deinterlace it using Hybrid as lossless (follow this guide), then upscale it using Gaia HQ AI. I think you will be getting much better results with less artifacts.
you can do a 10 sec’ preview in TVAI instead of attempting to upscale the entire video. you can do Gaia HQ 10 sec test (after you deinterlaced it in Hybrid) and Dione: DV on your original HDV file, 10 sec preview and see which one produces better quality.
you can use video-compare tool for that as well…
Seems the answer is changing the setting from “original pixel” to “square pixel”. Below are the images I got from two run-throughs. The first, the letter/pilarboxed image, is one using the “original pixel” setting. The second is the HDV 1440x1080 interlaced footage de-interlaced and up-scaled to 1920x1080 using the “square pixel” setting.
I do look forward to the day when Topaz comes out with a comprehensive manual. I’m sure it would describe in detail all the possible settings and the significance of each. I’m sure it would, in this case, tell us what the original or square pixels setting are and when they should be used.
