VHS-> DVD : Getting rid of grain in VHS tapes?

[Wile_E]

I'm capturing using Canopus ADVC-300, JVC HR-S9911U, and TMPGEnc encoder. What is the best way to get rid of grain/pixelation in the captured video? I've tried setting the Canopus sharpness control to softer, but it gets rid of some details.

[KenJ57]

I'm capturing using DataVideo DAC-100, JVC HR-S7800U (with TBC/DigiPure+R3) and Adobe Premiere Elements. The DAC-100 has a "wide and flat" video response and captures quite a bit of video noise along with great detail. I was instructed in this forum, first of all, to turn off the R3 circuit since it is intended for viewing but too agressive for capturing. Helped greatly!

There still remains a lot of video noise IMO, so I use the MSU Denoiser VirtualDub filter with very agressive settings to 'scape and smooth' my videos. This filter was designed specificly to remove noise and retain detail. My videos now look more like the images I get from Satellite TV. Only problem with a noise filter like this, is that it takes lots of time to process the clips. 13 minute clip takes about 3 hours on a 1.5 GHz P4 system. I like the results so much that I run the filter over night.

[lordsmurf]

TMPGENC NR filters

[guns1inger]

avisynth and convolution3d. For older tapes, the highest pre-set will probably be needed, but it does a damn fine job, and is faster than tmpgenc's filters. This is a basic script I used for a particularly grainy and well loved video tape (the tape was a good 10 years old, and was itself a dub to begin with)

LoadPlugin("g:\video\avisynth 2.5\plugins\Convolution3d.dll")
avisource("D:\CAPTUREs\your_avi_here.avi")
crop(24,4,-24,-4)
ConvertToYUY2()
odd=SelectOdd.Convolution3D (0, 32, 128, 16, 64, 10, 0)
evn=SelectEven.Convolution3D (0, 32, 128, 16, 64, 10, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()
AddBorders(24,4,24,4)
ConvertToRGB(interlaced=true)

[DVWannaB]

speaking of avisynth, I like Convolution3D filter a lot also, but recently I tried a combination of DeGrainMedian and FFT3DFilter and honestly, it crushes Convolution3D. Why? Because when set properly, they preserve more/better detail. Only drawback, is that they are slow. Convolution3D blows them away with sheer speed. But in the video biz, like they always say, "If you want better quality, you have to wait a bit longer to get it".

[guns1inger]

I'll have to check them out. Cheers.

[lordsmurf]

I never worry about speed unless it something unreasonable (12+ hours to encode a small clip or short movie). I let it work while I eat dinner, sleep, and eat breakfast. That's at least 10-11 hours there, if it needs it. Usually 4-6 hours is plenty, even on "older" CPUs in the 1.x or 2.x Ghz range.