frameserve FROM TMPGEnc?

[JamesB69]

Is there a way to frameserve from TMPGEnc to another program?

I do not really like its encoder, but its IVTC filter is the best I have found (the one in 2.5 at least) and I would like to frameserve to something else for encoding. I do not really have the disk space to save to an uncompressed avi, so I was wondering if I can get it to frameserve.

[adam]

Yeah it is possible. Set up TMPGenc how you like it then hit file/save project. It will give you a tpr file. Then Open this file in VFAPI Converter and it will let you convert it to an "avi file," which you should be able to load into just about any encoder that accepts avi input. The avi file it creates is just the frameserve loader. You don't need to leave TMPGEnc open or anything. Just use that avi file and your source will be filtered through TMPGenc and encoded through whatever you load the avi into.

Good luck, both TMPGenc's IVTC filter and VFAPI are pretty slow, and that's before you even get to the encoding

[JamesB69]

cool. I thought it was possible, from the days when I actually read about this stuff, but could not figure it out.

I have been trying to find a different way to IVTC, but TMPGEnc 2.5 has the option to select which frames are kept, which ones are not, or even allows you to specify a pattern. It's auto scan does well to usually with only a couple interlaced frames getting through in a 1 hour video segment.

Too bad they crippled V3....

Anyways, thanks for the quick reply man.


EDIT: is tmpgenc doing any encoding in the method you posted, or is it strictly kept to the filters?

[adam]

No encoding is being done. Just filters and resizing and such, whatever you set.

[mikiem]

I'm far, FAR from expert with avisynth, but might want to check it out as an alternative to VFAPI. VFAPI works in the PC colorspace, which can be cool -- works with Vegas where Avisynth won't. But using Avisynth the color space handling (& conversion if nec.) is *Much* closer to the original, don't suffer near as great a quality hit. Might want to check it out.

[adam]

Well avisynth is great and there are some IVTC filters for it that work even better than TMPGenc's, in my opinion, but it could only work as a replacement for the entire process he's doing. It cannot replace VFAPI as the frameserver for TMPGenc because avisynth simply cannot open TMPGenc project files.

[JamesB69]

well, I am capturing from a pci tuner (Theater 550, i've been capturing both MPEG-2 and Picvideo MJPEG), and I really don't know much (any) about colorspace. Will I be taking a quality hit by doing it my way?

THe reason I like TMPGEnc's ivtc is because i can launch and give it a pattern, and I can go to the frame where it changes and then alter the pattern. I havent found any other filter with that ability?

[mikiem]

"THe reason I like TMPGEnc's ivtc is because i can launch and give it a pattern, and I can go to the frame where it changes and then alter the pattern. I havent found any other filter with that ability?"

To tell the truth I don't know if any exist, not that I've come across -- which btw isn't saying all that much OTOH, don't know how critical that capability is. I've used V/Dub's version, & have used a number of adaptive filters in V/Dub, and as Adam points out, there are a number of great alternatives using Avisynth.

"Will I be taking a quality hit by doing it my way? "

My understanding is VFAPI forces a conversion to RGB. I used it a LOT with DGIndex, but then Donald Graft posted that he was considering dropping VFAPI support. The reason was that the color conversion, and how it was handled using VFAPI, was very inferior to using Avisynth. I altered my workflow, tried it, and personally feel the result was worth it. But as Adam posted, apparently there is no way to adapt it to your current workflow -- something I was not aware of. About the entire scope of my Avisynth knowledge has involved using *very* simple scripts.

As far as colorspace goes, there's a ton of info out there, so my advice would be to start at video help, then google & check the microsoft site depending on how in depth you want to go. A *VERY* basic intro would be to say that TVs are designed around a different method of storing color data then what you're used to with your PC. Video compression (mpg2, mjpeg etc.) captures that color data, and can store it in different ways -- some gets abandoned to reduce bandwidth/bitrate. On top of that, a video stream headed for your TV will *normally* not include the full range of colors used by your PC.

That all said, in case it helps... My ati aiw seems very tuned to mpg2 capture -- results *to my eyes* are better then picvid mjpeg. It might be fair to assume that the Theater 500, having hardware mpg2, might be even more tuned to that format, but everyone's eyes (& capture sources) are different, & mine are busy right now so you'll have to be the judge.

[manono]

Yeah, mikiem's right. If the encoder you'll be using accepts AviSynth script input, then it's a much better way to frameserve when compared to a VFAPI, which is slow, obsolete and inferior. There are 3 major IVTC's in use in AviSynth. You don't have to find the place where the pattern changes as you do using TMPGEnc's IVTC, as they adapt on-the-fly, so to speak. In addition, if there are problems, 2 of those IVTC's, (Decomb and TIVTC) allow you to tweak it in ways similar to that of TMPGEnc's IVTC. But it isn't usually necessary except sometimes with some difficult anime. Its colorspace conversions are the best. There's a bit of a learning curve with AviSynth, but if you do lots of encoding, it's well worth the effort.

[JamesB69]

I am trying this avisynth with decomb since it seems that Virtualdub will change to RGB and reduce quality(?).

What can I use to cut commercials and stay in the right colorspace? I am used to cutting with Virtualdub-MPEG2 and frameserving, but that evidently outputs RGB too...

[manono]

Hi-

Yes, if you use Full Processing in VDub and use the VDub filters at all, then you're changing color space. Use Fast Recompress in VDub (Video->Fast Recompress). It'll give you something like a 25% boost in encoding speed as well.

I don't capture, so I don't know the answer to that new question. Can't you edit the video (MPEG?) prior to encoding, using something like Womble or Cuttermaran?

[JamesB69]

I am trying tmpgenc MPEG editor for the commercials. It claims to be lossless.

Anybody know if it is staying in the righ colorspace?