I have a couple of deinterlaced ProRes videos that I encoded using Hybrid, sourced from Video8 tapes initially captured in VirtualDub. I'm looking to further process these ProRes files with Hybrid to share them with others in an h.264 encode.
Given that my input resolution is 720x480, I'm removing 20 pixels horizontally and 8 pixels vertically to account for headswitch and overscan, which brings the resolution down to 700x472 after cropping. I would then be resizing it to a target resolution of 704x480 for a proper 4:3 aspect ratio using the "Picture Resize" setting.
Here's my dilemma: Does resizing "up" from 700x472 to 704x480 technically degrade the video quality? Or would it be better to upscale it back to 704x480 instead? Any advice or insights would be greatly appreciated. I attached a screenshot to show what i am attempting to do.
Basically, any processing will technically degrade the video quality.
So, answering your title-question: Yes
Sorry, but what you write about aspect ratios does not make sense to me.
For further details about PAR&Co, reading
[INFO] About pixel aspect ratios,.. and the linked articles there might help.
If your source is not properly flagged and hasn't a 1:1 PAR, you should set the 'Input PAR' to the correct values.
If you want the minimal possible distortion, then set the output PAR you aim for and set just the width. (Do not disable 'Auto adjust'.)
What you are doing might be correct for your source, and I might just not understand you correctly.
Like I wrote: Yes, whatever you do you technically lose quality.
(using non-square pixels will technically additionally alter the video quality
)
Cu Selur
(03.05.2024, 21:13)Selur Wrote: [ -> ]Sorry, but what you write about aspect ratios does not make sense to me.
For further details about PAR&Co, reading [INFO] About pixel aspect ratios,.. and the linked articles there might help.
If your source is not properly flagged and hasn't a 1:1 PAR, you should set the 'Input PAR' to the correct values.
If you want the minimal possible distortion, then set the output PAR you aim for and set just the width. (Do not disable 'Auto adjust'.)
Let me see if I can clarify:
My original source is captured at 720x480, which has a "stretched" 3:2 storage aspect ratio (SAR). What I'm aiming for is to achieve a display aspect ratio (DAR) of 4:3 while also eliminating headswitching and overscan
in Hybrid. To do this, I'm cropping the image to 700x472 to remove headswitching and overscan. However, I'm concerned that resizing it to a higher resolution of 704x480 might negatively impact the image quality during rendering. Would resizing to 704x480 still be acceptable, or is there a better approach compared to what I did in my screenshot?
Or to make matters "easier", what values should I enter in the PAR and Picture Resize boxes within the Crop/Resize window to attain a 4:3 Display Aspect Ratio (DAR) considering I've already specified my desired crop dimensions resulting in a crop resolution of 700x472 in the Picture Crop box?
Are you sure Hybrid is the right tool for you?
PAR = DAR / SAR
If your DAR should be 4:3 with a given SAR of 3:2 one would use:
PAR = 4:3 * 2/3 = 8/9 which is a typical MPEG-2 4:3 PAR.
display aspect ratio (DAR) – the aspect ratio of the image as displayed
pixel aspect ratio (PAR) - width of a pixel compared to the height of that pixel
storage aspect ratio (SAR) - number of pixels in width divided by number of pixels in height
So if you crop to 700x472, your SAR is 1.483...
What you should do: (to minimize resize errors)
1. find the right PAR of your source, see the link I posted before, adjust the 'Input PAR', check the Vapoursynth preview. If your content aspect ratio wise looks correct in the preview, you got your PAR.
2. set the output par to whatever you want. If you want a specific storage aspect ratio (i.e. for DVD, Blu-rays,...) you should select width or height so that both are in that range.
3. set the output width or height to what you are looking for
4. if you need a specific storage resolution (i.e. for DVD, Blu-rays,...) enable Letterbox and specify it.
Assuming you use a decent player for playback, if you crop your content, so that no black bars exist and your input PAR and output PAR are set properly, you do not need to care about the DAR and adding black bars, since the player will care about it.
Cu Selur
my 2 cents... better to not resize anything if the issue are 2-40 pixels lost... just use letterbox (and a correct PAR to make sure a decent player can handle and resize on the fly the clip... leave black borders to maintain a standard resolution .. ) resizing is just losing quality and details in a bad upscaling (for a bunch of pixels) .. and losing a lot of CPU time...
btw if you need to fill entirely the TV screen with an old PAL/NTSC footage .. risizing is necessary but it's BETTER to keep the original clip without resize for future editing purpouses.
Quote:Does resizing "up" from 700x472 to 704x480 technically degrade the video quality?
yes it does, and even more than you would imagine.
if you had "inprinted" subtitles (like the ones i'm about to upload in another thread about vhs) in your video you would see this easily.
number of pixels doesn't even matter much, 1 pixel resize already makes a big differnece, especially in vertical direction, ie 480 lines in your case.
it's the skeleton of the picture that shouldn't be touched.
you can play with horizontal pixel count, though. usually a lot.
same thing on playback on modern screens: it's better to watch in 1:1 pixel mapping, ie 1080 looks best on 1080 screen, not on 4k screen which needs upscaling.
in that case avisynth/vapour will usually beat tv's resizer/deinterlacer.
offcourse, these days you can probably do qtgmc for sd content in realtime, ie connect pc to tv.