Noticed this in 22.5.0
So my source video is a HD ProRes 422LT in Premiere, but when exporting the same file: ProRes 422LT out of Media Encoder-with no effects or differences to the timeline the export time says it'll take like over a day. It used to be in previous versions that it would be a straight copy protocol of the file. It just seems to be in limbo when exporting out of Media Encoder, it's super fast and works as normal when exporting out of Premiere Pro.
Tried reinstalling all Adobe software but same problem.
This is very aggravating exporting one at a time through Premiere and not queuing everything through Media Encoder.
Exporting the same settings

We can 'delete all empty tracks', or we can delete 1 specific track, but there is currently no way to delete multiple, but not all tracks. The most logical way would be to delete all activated target tracks, as this could be done from a single keystroke, most logically the backspace key. I'm surprised really that there is already a key command called 'delete tracks' that's written in plural but actually just brings up a window where you may choose one potential track to remove, or just all empty tracks, I think this new function should take the place of that command in the list of mappable key commands as it would better match what one would expect 'delete tracks' would do.

The new UI is unfortunate, and terribly inefficient. If this is the path forward that Adobe is taking, please take into consideration that many professional video editors use 2 screen setups.
Your new IMPORT, and EXPORT UI’s do not have docable windows. This makes it more difficult to set up for a 2 monitor system.
Even if you do make adjustments to the UI, they are not retained by the program. The EPORT page puts the split of my 2 monitors directly in the middle of the output preview. If I shift it, it will be reset next time I go into the EXPORT page.

I’m not quite sure what the though process was with some of these new UI element. Specifically the “IMPORT’, “EDIT”, & “EXPORT” windows. These are infuriating. They server no purpose.
Starting a new file use to just start. From there the user could attack the project in whatever way they wanted. This IMPORT page doesn’t allow the use to move forward without predefining a project name, and using a dropdown setting that is WEAK in comparison to the operating systems basking file saving explorer window. SOOO MANY CLICKS! Why?
So many GIGANITC folder icons. Why?
All of this is useless. CTRL “N” and CTRL “I” did all of this in 2 shortcut clicks.
You can fix this by simply allowing the user to skip this page and go right to the EDIT page. This would allow us to skip past this useless page, and allow anyone who does see value in it to keep using it. But as it stand it’s EXTREEMLY FRUSTING to use this, and it the first thing you have to do when opening the software.

In all older versions, command M opened ME immediately. Now it opens the Export window, makes you click your way to ME, then slowly opens ME, then transfers data to ME, then you have to set up your files in ME (which you used to be able to automate in preferences, set destination etc - no longer works, check it) and then your presets don't carry over to the next encode, which they used to.

What a disaster! You realize that probably 90% of your users render using ME so they can continue editing, right? So why did you make it so clumsy and clunky in this new release? It's a step back of about 5 years.

I love the new closed caption capability - thank you.
However it has a significant problem: it does not correctly calculate the time to STOP displaying the caption.

So if you have a narration audio track with short chunks of narration, the captions continue to display until a couple of seconds before the next chunk begins - resulting in an absurdly long time on screen.

See the sequence illustrated in the screenshot attached. Some of the captions are stretched for a very long time - eg 34 seconds to read 15 words.

Suggested fixes in order of complexity:
- Stop all captions after a maximum of n seconds
- Give the user a slider (like the one for minimum duration)
- Calculate the duration based on number of words (with the user having a slider to adjust anticipated reading rate).

I do media analysis on YouTube, which requires me to use footage from movies & television to complement the narration. Scene edit detection has been great for this, as I can just import the files into my workspace, and premiere will automatically chop them up into subclips before depositing them into a folder for later use. It's honestly been a game changer for me.

That being said, there has been a bug that has been bothering me for quite some time, and to date, I am yet to find a solution.

For some reason, whenever I drag a SED subclip into the timeline, it always adds on the final frame from the previous shot during the import process, meaning I have to cut that first frame every single time or the viewer will be bombarded with flash frames. Sometimes this isn't a problem if I'm intending on shortening the clip even further anyway, but there are points where I do just want to drag & drop clips into my timeline, and it's a giant pain to have to chop it down every time.

This first frame doesn't appear in the subclip preview & this problem is also not present when I go back to the original chopped up file. This is a problem exclusive to the import process.

I've been searching for years at this point for a solution, but no one else has either noticed this issue or has been bothered to bring it up, and my patience is wearing a bit thin. I waited a little bit to reach out to see if any future updates would fix the problem, but it's become clear that this is not on Adobe's radar, so I figured it was about time I draw some attention to it & see if a solution already exists.

I've also attached a video demonstrating the problem, in case my description wasn't clear.

I’ve noticed what appears to be a bug in how Premiere renders H.265/HEVC footage after applying certain effects. From what I can tell, if an effect is not natively GPU accelerated, the image will be degraded by color noise/artifacting and perhaps a color shift. It essentially looks the same as if using MPE Software Only.

Effects such as Warp Stabilizer don’t cause this, as I don’t think they are in the same processing pipeline, but I have found that the following do:
Camera Blur
Neat Video
Color Balance
Flicker Free (older non-GPU acceleration version)
Luma Key
Posterize Time

Out of these, Neat Video’s noise reduction plugin is the most relevant to my workflow – but I wanted to show that this bug applies to both native and 3rd party effects. And while it’s a bummer that Neat Video for Premiere isn’t really usable with HEVC footage, I found that if I dynamic link to Ae and use Neat Video there, the issue isn’t present.

I’ve also included a couple of still frames for reference – notice the blocky color artifacts on the second image, especially in the purple areas to the left and right sides of the frame. The first image just has Lumetri applied, while the second also has Camera Blur but with the percent blur set to 0 (saturation has been boosted in both to help show the issue).

Lastly, I have observed this problem on a variety of HEVC footage (422 10-bit, 420 10-bit, and 420 8-bit) and two different workstations (specs below).

System 1
Windows 11
Premiere Pro 2022
i9-12900K
64GB DDR5 4800MHz RAM
RTX 3080
Multiple NVMe SSDs (500GB–1TB)

System 2
Windows 10
Premiere Pro 2022
i9-9900K
64GB DDR4 2666MHz RAM
GTX 1070
Multiple NVMe SSDs (500GB–1TB)