Punched-in footage exports blurry to lower resolutions

Hery Rach, just wanted to check in to see if there is any progress on this acknowledged bug. Googling around I found some other users stumbling upon it as far as 4 years ago, and the culript probably being "Use Maximum Render Quality" being completely broken for multiple years, at least on Mac.

Just to make another example on how this happened again in a real world scenario: customer needs corporate video (normal 2160p sequence) for an exhibition ledwall, but supplier needs file at the exact resolution of 840x672px. Below what happens exporting directly to a lower resolution, vs exporting to the composition resolution and then re-encoding in Media Encoder to a lower one.

Also notice how different the shadow around the title looks, definetly something funky going on when exporting to a different resolution in Premiere. You'd think you'd get the same output, just at a different resolution?

Sorry MyerPj, but you probably didn't carefully read the OP nor the rest of the thread, and I'm just tired of re-explaining it again.

Now at least the bug seems to have been "aknowledged" after 1 year and 4 months from first reporting, after having been played down multiple times by multiple users who for some reason just preferred criticising the workflow thinking it's my first day on the software, instead of actually looking into the issue.

Thank you to all the devs who investigated this. Open to more testing if necessary, being now updated to the latest Apple Silicon (M4 Max) and OS. Issue still there.

Scale to Frame Size is an option you shouldn't use, as has been said above.

The only reason I can think of to use it, is if you have a giant still image, and you're machine is just totally bogged down by it, scale to... would throw away the extra pixels (resolution) and allow your machine to run better.

Barring that, I can't think of a reason to keep it around, other than it's been there for a long time.

Use Set to Frame size which is now called Fit to Frame in v25, and also comes with a new option Fill Frame, which creates a bit of a crop (the pixels are still there) on the top/bottom or sides depending on your aspect ratios.

I can't replicate the behaviour you're getting. In your post, you mention using a setting that would guarantee getting the results you are reporting.

Hence my comments.

If you are getting this without using Scale To, in both the Preference and on the sequence, then that would be "wrong" behavior by the app.

I'm just practical.

Those are words you're typing on your computer.

Reality is whatever is happening in the test project and screenshot I attached, which still nobody really tried to replicate on their system, and whatever is happening to countless other users in the other topics I linked above.

Reality is this clip just exported from that same test project, where left side is on "Scale to" and right side is on "200%", using ProRes 422 default preset with FHD resolution selected.

Reality is encountering this issue for over a year, multiple times every week, just because I'm trying to use a really basic feature of Media Encoder to export a clip to multiple output resolutions.

I don't wanna sound rude but I've been ridiculed one time already because nobody really tried to replicate a clear-as-day issue, perfectly reported TWO times using the provided template with steps to reproduce and system configuration, and even attaching a ready-to-go test project in an attempt to improve the software for everyone. It always ends in a lot of discussion about technicalities but nobody really gets to what happens in reality.

No. Just .... no.

Scale to throws away your original resolution as an INTENDED thing,  as that was originally designed back when computers had like 16GB of RAM max. And struggled with playing back 1280x720 ProRes sequences.

Once thrown away, you cannot get the data back.

Using Scale To, starting with UHD to 'normal' 1920x1080, your image is ONLY 1920x1080. Period. There is no further data available for the computation. Including at export.

Using Set To, starting with UHD to 'normal' 1920x1080, you still have all the original pixels available for any further scaling work.

That's reality.

I'm sorry Neil, but the core of the issue is not even what the difference between "Set to" and "Scale to" really do: the issue is that -whatever- they do, it's not reflected correctly in the preview window VS what happens during export. I could edit all day, having perfect sharpness in my preview window at whatever size and/or resolution I have set it at, but whenever I export to a resolution that is LOWER than the composition resolution all the clips which have "Motion > Scale" in the EPC at any value above 100% are blurry.

I get what you're trying to get at with the difference between "Set to" and "Scale to", but in reality however they operate should be negligible: using "Maximum render quality" should still bring out all the available pixels during export. If I framed 1080p worth of pixels on a 2160p composition and export a 1080p clip (or lower), even taking into account any loss from scaling algorithms, I should expect close to 1080p worth of detail into my export. But as you can see from my test project, half of the frame gets RAMMED (the half which has been scaled with "Motion > Scale") while the other half (that used "Scale to") keeps detail.

This is a big issue, because as you said in your last post, I cannot use "position and Scale settings" for any clip if I intend to export some versions at lower resolutions.

Set to and Scale to are technically settings in the Preferences, and can also be invoked from sequences IIRC.

Both allow for dropping clips onto a sequence and automatically adjusting to fit the sequence framesize.

But Scale to rasterizes the image to the sequence. Any further scaling is done from that raster image.

Set to adjusts the image to the framesize but keeps the original data. So any further scaling is computed from original image data.

If you are simply using the position and Scale settings on a sequence in the ECP, that also works with original image data.

How am I supposed to use scale to framesize if I need to punch-in footage to just about 120%? Or If I need to do a slow zoom in, let's say from 100% to 120% using keyframes?