Please update comparision :)

[Dee Mon]

As I already said here many times, super resolution by definition is a method of increasing resolution, i.e. upsampling (upsizing, upscaling, uprezzing - call it as you like). If the resolution stays the same, it's not SR at all. If you tell VE to keep original resoluton it will just copy the video without changes.
Blending frames without changing resolution is just a method of removing noise. This is called temporal denoising and a VirtualDub filter for this task has been around for quite a while, although it's not motion compensated, so the quality is lower.

[MC87]

As I already said here many times, super resolution by definition is a method of increasing resolution, i.e. upsampling (upsizing, upscaling, uprezzing - call it as you like). If the resolution stays the same, it's not SR at all. If you tell VE to keep original resoluton it will just copy the video without changes.
Blending frames without changing resolution is just a method of removing noise. This is called temporal denoising and a VirtualDub filter for this task has been around for quite a while, although it's not motion compensated, so the quality is lower.

Yes, sorry - I remember I asked you what would happened if Video Enhancer is set to output at the original resolution.

If VE can only do 2x SR scaling if I am currently trying to upsize a 1200x600 video to 1080p, would it be better to do the normal 2x upscale to 2400x1200 or should I scale it to 1080p? - if I do this will the SR algorithm resize to 2400x1200 and then downscale it to 1080p? or will the SR process itself be done on 1920x1080 frames? (scaled up from 1200x600).

[Dee Mon]

Our SR method doubles the resolution, so if your input is 1200x600 and you ask for 1080p, SR will upsize the input to 2400x1200 and then resample down to 1920x1080. This is done automatically when output size is greater than input size but not exactly 2x.

[MC87]

Our SR method doubles the resolution, so if your input is 1200x600 and you ask for 1080p, SR will upsize the input to 2400x1200 and then resample down to 1920x1080. This is done automatically when output size is greater than input size but not exactly 2x.

Ah ok

Also is MSU smart sharpen any good (on default settings) - I was thinking of using it as the output is a bit blurry. (I would put this in the filter chain after SR resize, not before - correct?)

[Dee Mon]

Yes, sharpen filter should be used after SR, not before.
Don't remember how good is the MSU filter compared to other sharpen filters.

[tritical]

Thanks for putting up the high resolution videos.

[Arkamond]

What about Topaz enhance filters? I dont see those in the new comparison.

[Dee Mon]

Look at Instant HD. Topaz is exactly the same engine. (if I recall correctly, need to ask SeeingMole who performed the measurements).