Log in

View Full Version : VDub & AVS read errors


Piper
18th July 2003, 14:35
Recently I've started encoutering AVS read errors when running multiple jobs in VirtualDub 1.5.4. It will always do the first job correctly but fails after starting another job. Somtimes it'll fail on the first, sometimes the second.

I'm using my own VBLE compressed 640x480 VDub capture source. Before setting up the jobs, I do basic editing and save the processing settings.

I've tried compressing with the new VCM and the various DivX 5.0.5's, all produce the same result. My #1 suspicion is the VBLE codec. #2 is VDub itself.

In either case, here's my script. I'm using AVISynth 2.5.2.

AVISource("C:\capture.avi")
Limiter()

SeparateFields()
Convolution3D(preset="movieHQ")
Weave()

FieldDeinterlace(blend=true,dthreshold=4,threshold=8)
BicublinResize(352,480)
UnDot()

Is anyone else experiencing similar issues? When running multiple passes, this gets to be a real pain.

Wilbert
18th July 2003, 15:39
I can't help you with your problem. Do you have the same problem with Huffyuv?

I don't understand your script. I guess you have a ntsc source. You should deinterlace first before applying any smoothers, and not after!

Is your source telecined, trully interlaced or neither?

sh0dan
18th July 2003, 15:57
Try using Bicubicresize instead of Bicublin - it's the only filter that isn't widely used.

What errors are you getting, or what happends?

Piper
18th July 2003, 19:33
@sh0dan

I'll try a variety of different resizers and see if that makes any difference, thanks. I don't have the exact error handy, but it looked like a generic AVS read error. The next time I come across it, I'll post it here.

@Wilbert
I did have similar problems with VDub 1.5.2 with Huffyuv, yes. Those went away when I moved to 1.5.4. I've only just started experimenting with VBLE which is why I immediately suspect it or VDub as the source of the problem.

Yes, the source is NTSC, pure interlaced. I placed the smoother at the top of the script as to (albeit theoretically) retain more detail than applying after Deinterlace/Resize. Other than this, what about the script don't you understand?