With that zoom factor, maybe better to produce DVD's instead of Blu-rays.

This shows (in order) crop, crop NeatV5, crop TopazVEAI

Neat does the best at getting rid of the noise you don't like, Topaz does a good job at creating extra detail but faces often not so good, this is no exception. Topaz does not remove that area of artifacting .

Regarding choice of BD format, I see no reason to go from my 25p source to 24p with the consequential audio pitch change if no frame changes are made, or the consequential video degradation if frames are interpolated or dropped.

There is no 1920x1080-25p option in Vegas, but there is a 1920x1080-50i option, which I have chosen. With smart resample, Vegas merely makes the second field in the frame the same as the first. You can see this by loading the 50i video into Vegas, setting the project properties to 50p, and then stepping along one frame at a time. With true 50i video you will see 50 distinct frames per second, but with 25p source you will see 25 pairs of distinct frames per second, where the two frames in the pair are identical. How TVs display interlaced video on a progressive screen and what deinterlacing they do is a mystery to me, but I think I am on pretty safe ground in assuming that my 50i it will look the same as if it had been 25p.

I chose the zoom value of 5.5 or whatever it is to exclude all members of the choir to our left of the compere. I thought that preferable to the increased artifacts of zooming.

The compere is only of secondary importance to the choir, so lowering the resolution of my final video to give uniform resolution throughout seems like a waste. Feel free to differ with me on that point, of course. If a TV news broadcast shows old SD footage during a HD transmission, they don't drop the whole news session back to SD to keep the appearance uniform.

todd-b, thanks for your post. I think NeatV5 looks good enough and better than Topaz.

There is no 1920x1080-25p option in Vegas, but there is a 1920x1080-50i option, which I have chosen. With smart resample, Vegas merely makes the second field in the frame the same as the first. You can see this by loading the 50i video into Vegas, setting the project properties to 50p, and then stepping along one frame at a time. With true 50i video you will see 50 distinct frames per second, but with 25p source you will see 25 pairs of distinct frames per second, where the two frames in the pair are identical. How TVs display interlaced video on a progressive screen and what deinterlacing they do is a mystery to me, but I think I am on pretty safe ground in assuming that my 50i it will look the same as if it had been 25p.

Even if you start with 25fps, you're still not going to reliably get the same or better quality going from 1080i50 to 1080p25, as you're relying on the TV's (or player's) deinterlacing to not suck... Smart deinterlacing only works in VEGAS when your SOURCE is interlaced, if you encode at 50i, 50i is what goes on the disc... you send 50i to the TV, so there is no deinterlacing happening until it gets to your TV, and the quality of those deinterlacers can vary dramatically. So, you will always get better image quality going from progressive scan to progressive scan than you will going from progressive scan to interlaced.

Your end result will look better if you have 50 full frames showing (25fps source with doubled frames) at 720p50 vs. 50 half frames that the TV then has to de-interlace into 25 full frames at 1080i50, because as I mentioned earlier, some TVs and players deinterlace better than others. For instance, if the TV chooses to interpolate the fields that should be taken from the next frame instead of just taking the empty fields from frame 2, you've thrown away half your resolution right there... and filled in the blanks with the TV's best "guess" of what should have been there... and few TVs and players give you control over the deinterlacing method.

Interlacing artifacts:

You also might want to experiment with setting the resample mode to "optical flow" before encoding to 720p50, and see if letting VEGAS add in the missing frames looks better than downsampling to 25fps by doubling frames from a 50i or 50p source. I've had some pretty good results using it in the past.

fr0sty

I think what I should do is first render my 25p UHD project to 50p 1920x1080 HD, with the clip properties set to "disable resample". The output file will then have duplicated frames in order to double the frame rate. If the clip properties are set to "smart resample" or "force resample", the additional frames will be interpolated or merged versions of the two frames either side of the new frame, which I don't want. Now, when I render the 50p video to 50i, each field will pick up the appropriate line and I won't lose vertical resolution. I will now try to verify all that.

This takes care of you up to the point where you are still relying on the playback device or TV to choose the proper deinterlace mode (blend fields) to deinterlace the resulting 1080i clip that you render from VEGAS... if it chooses "interpolate fields" instead, you're actually losing quality. This is why I suggested going with 50p at 720p instead, it stays within blu-ray spec, and you aren't at the mercy of the deinterlacer on the player or TV choosing the right mode to get the effect you want. In that case, it just doubles each full frame so you still see 25fps, even though it is displaying 50.

I'd do a test with Frosty's method as well and see how it works on a TV or two. I doubt the 1080p vs 720p difference will be as noticeable as how deinterlacing is done.

Hmmm, it looks like the wrong pair of fields are getting associated to form a frame. I need time to explore further, including 720p.

@PeterDuke

... and he came running in from left field with another hair-brain idea.

I think I'd let the aspect ratio go below 16x9 before I let the vertical dimension go below 1080 (or 720).

Especially, in this day and age...

P/S

I always set my key frames to integers for horizontal and vertical dimensions as well as horizontal and vertical location in Pan/Crop. I don't know what algorithm Vegas uses, but, whatever it is, subpixel sizes and locations probably won't make the output any better.

I don't know what algorithm Vegas uses, but, whatever it is, subpixel sizes and locations probably won't make the output any better.

Depending on render quality setting "good" or "best", Vegas rescales "bi-linear" (good) or "bi-cubic" (best). Bi-cubic is the best resizing algorithm available in Vegas.

@PeterDuke Here is my take, based on my idea a few posts up. This project used two intermediates -- one a lossless 2x crop from 4k to 1080p, next a Vegas Upscale at 2.75x, and finally a delivery render at 1080p, which is on the right. Your upload is on the left for comparison.

Probably my eyes abandon me, but only a little less noise, I see no improvement of details due to the applied Upscale FX.

As said, just less noise.

An alternative solution to make an intro without zooming in too much.

I tried a Vegas only effort, using Vegas Denoise + Vegas upscale. It looks like this is the type of noise Vegas Denoise was designed to fix (iso noise) , however it also added vertical lines that made it unusable. This is a Denoise bug, not upscale

john_dennis  and 3POINT

Your suggestions remind me that we see portrait video from mobile phones on landscape TV all the time, so maybe most people wouldn't mind. The only problem is Peter Duke would😝.

Musicvid, your suggestion is interesting, but I think it would still need a dose of Neat Video or similar.

john_dennis  and 3POINT

Your suggestions remind me that we see portrait video from mobile phones on landscape TV all the time, so maybe most people wouldn't mind. The only problem is Peter Duke would😝.

Then Peter Duke should put some more effort in his project while recording instead of trying to correct his amateurish recordings afterwards in post...