Copy link to clipboard
I'd like to know the difference between cropping and scaling a video clip directly on a Premiere Pro timeline, versus doing so in the Adobe Media Encoder export window?
I am converting DVDs of VHS source material to H.264 files and want remove the distortion at the bottom (head tearing), and on the sides. I scaled each clip up 104%, which successfully cleaned up the bottom and sides before sending the project to Adobe Media Encoder for export.
The 'Source' tab in Adobe Media Encoder has several options for cropping, while the 'Output' tab's 'Scale to Fill' seems to do the same thing as scaling in Premiere.
Is there a fundamental difference between doing these operations in a Premiere project versus doing them in Adobe Media Encoder?
Copy link to clipboard
both are valid possibilites. If you want a quick transformation or prepare your ingest data quickly AME with Crop and Scale automatically to your desired output size is the best choise. If you need some additional work or effects on it or require some async scaling or free transformation of the source Premiere is the better choise.
If you want do multiple batch processing, e.g. covert a lot DV Tapes, DVD, VHS into another format, I would definitive build your custom preset, including the crop and scale option, and us this every time. Normally your input has the same black bars and that should work. If the black bars changes you need to redefine your crop by hand.
Hope that helps,
Thank you SvenS,
Will the video be adversely affected by the scaling, whether done in Premiere or in AME, since I will be converting from 720x480, 4:3, NTSC non-square pixels to square pixels with a yet to be decided aspect ratio and resolution? It has been recommended for me to use 720x534 as the exported resolution.
Side note: Is there a big quality difference between exporting as a 4:3 SD file and upscaling the export to something such as 16:9, 1280x720 HD? The only difference I've noticed on past exports is the pillarboxing on 720p files, which is present anyway when viewing 4:3 in a player using fullscreen mode. Am I close to being correct?
Copy link to clipboard
Drop the 720-by-480 non-square pixel source into a 640-by-480 square pixel Sequence in Premiere Pro and then scale to about 104% to do what's called "crop for overscan" and then send that to Media Encoder to create H264 MP4s. Use the Match Source - High Bitrate preset.
Drop 720-by-480 non-square pixel source into a 1280-by-720 square pixel Composition in After Effects and then apply Detail-preserving Upscale Effect. Use the "Scale to Comp Height" button for this Effect and then increase the Scale percentage until the skew is cropped and then send that to Media Ecnoder to create H264 MP4s. se the Match Source - High Bitrate preset.
Thank you Warren,
Unfortunately, all of the sequences are already completed with edits, adjustment layers, remastered audio tracks, and video noise reduction applied, so I'm pretty much locked into the native 720x480, 4:3, non-square pixels sequences. That said, what resolution and aspect ratio can I use for exporting the completed sequences? I've done some test exports, upscaling to 1280x720, 16:9 and they look pretty good. I have read comments that baking in the pillarboxes for 4:3 to 16:9 conversion is not recommended. Why is that? Whether I export 4:3 or 16:9, there will still be black bars on the sides when played in fullscreen mode in Windows Media Player, VLC, etc.
You can conform to 640-by-480 square pixel before editing or after editing. Or you can conform to 1280-by-720 square pixel before editing or after editing.
Pillarboxing is only important if you want to control how the original 4-by-3 frame aspect ratio appears within 16-by-9.
Upconverting to 960-by-720 or 1440-by-1080 instead of 1280-by-720 is a fine way to go and probably better if the movie might be played in a free-floating window instead of full screen.
It's extra work, but you could upconvert a "_full" version (full frame 4x3) and a "_wide' version (widescreen 16x9). So, if your edited 720-by-480 movie file is named "family_wedding_dv.mov", you could create a 960-by-540 version and name it "family_wedding_full.mov" and another 1280-by-720 version named "family_weeding_wide.mov".
We see this with DVD-Video where there's a full frame version and a widescreen version. Sometimes they're separate DVD-Video discs sold separately and other times it's a 2-sided DVD-Video disc with full screen on one side and widescreen on the other.
Regarding pillarboxing, I always watch videos in a fullscreen window, so black bars generated by a player or baked into a 16:9 file don't bother me. These videos are going to others though, and I don't know how they will be viewing them, other than they want to be able to watch them on computers, TVs and phones.
In another thread, Ann Bens recommended exporting as square pixels, 720x534, and someone else recommended 720x540. How do these compare with your recommended 960x720 or 960x540?
Also, does it matter which codec is used, and should I be using the "Main" or "High" profile?
I do appreciate your time in helping me with all of this.
Let me see if I understand it correctly: You have DVD, VHS footage in (SD) 720x480, NTSC 4:3, none-square (480i - FullFrame https://en.wikipedia.org/wiki/Standard-definition_television#Pixel_aspect_ratio ) and you want to convert it to some format that can be correctly viewed and displayed on a 'modern' device, correct?
There are often 4 factors that can result in issues or by what reason the device struggles:
As I understand your issue is the first one, that the playing device cannot handle the PAR of your current file correctly and you want to convert it to a 1:1 (square pixel) format. So the minimum would be: Using h.264 format, keep the original settings 720x480 and just set the PAR to 1:1, best set it to progress too and you should be done. This is then 4:3 with NTSC square pixel and progress and should play on all devices. On the device itself you can decide the scaling option. The benefit of this 'smaller' size is, that you keep the lines (height) the same as the origin and the interlace to progressiv frame could be more precise.
So I would suggest a H.264 - Match Source (Medium or adaptive Medium). The Target bitrate is responsible for the quality and for the generated File size. A simple comparison is: If you original video is e.g. 1GB, you output should be in the same order too to have a similar quality. Smaller File sizes, you will lose quality; bigger file sizes only could help in some cases, e.g. upscaling. It depends on the operation and you need to compare it visually. For this case here you can play with the video->Target Bitrate to match the estimated output file size with your origin file size.
As another option you can set the output size to 16:9 and decide to the different scale options you want. You can use the 853x480 and keep the number of lines the same,upscale too if you want (HD/Full HD) or let the player device doing the upscale later. That you need to check for yourself to see what is looking better.
In general AME would be your tool of choise for this work. I would suggest to open the export settings and make a comparison in the Preview: Source and Output while changing the settings. That helps you to easily see, what your settings will do to the final output.
Hope that helps.
Using h.264 format, keep the original settings 720x480 and just set the PAR to 1:1, best set it to progress too and you should be done. This is then 4:3 with NTSC square pixel and progress and should play on all devices.
Hope that helps.
By @Sven S
A quick technical note: The pixel aspect ratio (PAR) value provides the scaling instructions within Adobe applications (Photoshop, Premiere Pro, and After Effects) such that 720x480 will always appear proportionally correct when added to square pixel documents, Sequences, and Compositions. As such, we should avoid rendering or exporting 720x480 as square pixels (the picture will appear slightly too wide). Unless we know the player knows how to correctly display 720x480 non-square pixel footage, it's a best practice to conform it to a proportionally correct square pixel frame size (640x480, 648x486, 720x540).
Yes Sven S, you understand it correctly. I will be exporting (SD) 720x480, NTSC 4:3, non-square (0.9) sequences to H.264, square pixels (1.0), deinterlaced files for viewing on a wide range of devices, and perhaps uploaded to YouTube. Note: I've never used Stretch to Fill for reasons you've touched upon.
Any of the following are fine for conforming 720-by-480 DV-NTSC for viewing on a computer or mobile device as 4:3:
960 x 540 is half HD 16:9.
I'd use the "Match Source - High Bitrate" H264 preset. It uses the Main profile.
Okay Warren, tell me if any of the following is incorrect:
The native 720x480, non-square pixel clips have been scaled to 104%–the "crop for overscan" you had mentioned.
I chose the H.264 format in AME, which gave me the DV NTSC preset by default–there is no "Match Source - High Bitrate" listed anywhere. I changed the resolution to 720x540, then the field order to "Progressive" and the aspect to "Square Pixels (1.0); Profile: Main, and Level: 3.1. I set the Target bitrate to 6 Mbps and the Maximum to 12 Mbps–the quality of the source footage is quite good, especially for Hi8 and VHS sources.
Since I scaled the clips by 104% in the sequences, do I need to add a 4:3 crop in the Source tab? Adding the 4:3 crop shows 705x480 pixels in the crop window. Is that even revelent?
Should I leave the Output as Scale to Fit? I thought that since I already scaled the clips on the timelines that I wouldn't have to Scale to Fill on export. Please let me know if anything needs correcting.
Thank you again.
I have always taken care or the format changes in either After Effects or Premiere Pro and then sent the resulting square pixel Composition or Sequence to Media Encoder.
I haven't tried doing it with source footage directly in Media Encoder.
I'll try your steps later today and see it they'll result in something that'll look good.
Just re-read your steps.
Scale to Fit is good.
Do I need to apply a crop under the Source tab in Media Encoder?
When I apply a 4:3 crop to the Source, and then Scale to Fit in the Output, there is a slight difference in the Output window when compared to not adding a crop at all.
<With no crop applied to the Source, there are very slight black bars on the top and bottom in the Output.
<With a 4:3 crop applied, there is a slight black bar on the right side of the Output.
The Output is completely filled in both instances when Scale to Fill is applied.
Note: The source clips were scaled 104% in Premiere.
I have attached screenshots to what I am referring.
If the picture is edge-to-edge in the Premiere Pro Sequence (no black on the sides) you should see thr same tbing when you go to export.
I just read some comments elsewhere stating that 720x480 should not be resized to 720x540, as changing the vertical dimension can have adverse affects on the deinterlacing–only changing the horizontal dimension is recommended.
Can you shed some light on this, and why some may be led to make such claims?
We can make a strong argument for confirming 720x480 to 640x480, but the picture will hold up to being changed to 720x540 well enough.
For any larger frame size from 720x480, I would use After Effects Detail-preserving Upscale or Topaz Labs Video Enhance AI.
Should I mention other software upconversion options available in Photoshop, Compressor, or Resolve? Or hardware available from Taranex or Panasonic?
After numerous tests, I settled on exporting the 720x480, non-square pixels, interlaced sequences (scaled 104% to clean up the borders) as 640x480, square pixels, progressive in Adobe Media Encoder. In AME, I chose crop to 4:3 with the output being Scale to Fill. I did notice the video being slightly less sharp when toggling between the Source and Output windows. I figured this may have been caused by deinterlacing and the conversion of non-square to square pixels, as shown in the Output.
Note: I noticed from my tests that scaling the clips on the 720x480 timelines adds nearly three-times the amount of time required during exports.
I've been given a couple more DVD-Rs that also need to be converted for viewing on computers, and I want to try your workflow of using 640x480 sequences from the very start. Do I simply set up a new sequence using "Custom" settings: 640x480, square pixels, and progressive, then drop the 720x480 clips onto the Timeline, choosing "Keep existing settings"? After that, should I use the Crop tool or Opacity mask to clean up the borders? These DVD-Rs also contain VHS footage, and thus have distortions on the bottom and on the sides that I want to remove.
Regarding the other software options you'd mentioned, I have no experience with using After Effects for this type of work, but I would like to explore that avenue in the future. I have looked at Resolve for color correction/grading, and also Topaz AI.
In the past, I've been very happy with the results from Premiere Pro with the Neat Video plugin.
Thank you again for your time.