WorBry
1st July 2007, 10:55
For Pal DV to MPEG-4/AVC conversion, I have been in the habit of resizing (after deinterlacing) directly to 720x544 (Edit:720x528 :rolleyes: old habits die hard) (720x400 for widescreen) with LanczosResize, and then denoising.
Recently I received advice that for 16:3 anamorphic DV (720x576) it is preferable to first size to 'full' square pixel resolution (1048x576) before down-scaling, to avoid introducing additional artifacts, unless "you have very good software that can scale the image without distortion".
Would appreciate the comments of our resident experts on this.
If this is considered sound practice:
a) Should the same be applied to 4:3 DV i.e. first size to 788x576 and then 'downscale' to 720x544 (Edit:720x528 :rolleyes: again )
b) Would Bicubic be a better choice for initial sizing?
c) Are there any particular benefits, in terms of image quality, in denoising before down-scaling (I know it will be slower).
I'm processing in YV12
Thanks alot.
Recently I received advice that for 16:3 anamorphic DV (720x576) it is preferable to first size to 'full' square pixel resolution (1048x576) before down-scaling, to avoid introducing additional artifacts, unless "you have very good software that can scale the image without distortion".
Would appreciate the comments of our resident experts on this.
If this is considered sound practice:
a) Should the same be applied to 4:3 DV i.e. first size to 788x576 and then 'downscale' to 720x544 (Edit:720x528 :rolleyes: again )
b) Would Bicubic be a better choice for initial sizing?
c) Are there any particular benefits, in terms of image quality, in denoising before down-scaling (I know it will be slower).
I'm processing in YV12
Thanks alot.