Page 1 of 1

Re: telecide and y0/y1 usage

Posted: Sat May 24, 2014 8:43 am
by admin
Can you give me a source sample that displays your issue that I can use to experiment with?

Re: telecide and y0/y1 usage

Posted: Sat May 24, 2014 8:48 pm
by Aleron Ives
To my knowledge Don doesn't operate an FTP drop. You have to provide your own hosting.

Re: telecide and y0/y1 usage

Posted: Mon May 26, 2014 2:22 pm
by admin
Will have a look tomorrow.

Re: telecide and y0/y1 usage

Posted: Wed Jun 04, 2014 8:19 am
by admin
First question. You said it is 3:2 pulldown material. But when I look at your sample it is progressive with a dup every 5 frames (disregarding the scroll). So Telecide() should not even be applied.

Before continuing I will wait for your clarification.

Re: telecide and y0/y1 usage

Posted: Wed Jun 04, 2014 1:14 pm
by admin
Even given that, you will have a big problem eliminating the dups while keeping the scrolling smooth. There is no good solution for that.

It may be true that Decimate requires equivalent range commands, but you haven't really shown that with a sample. I'd be willing to implement that if you have a sample that clearly demonstrates the need.

Re: telecide and y0/y1 usage

Posted: Wed Jun 04, 2014 4:11 pm
by admin
Great. Thank you for the update.

Re: telecide and y0/y1 usage

Posted: Wed Jun 04, 2014 10:41 pm
by admin
MrVideo wrote:I need to look into the forcefilm option for 4:2:0 sources. That was not usable here, as the source was 4:2:2.
Huh? What exactly are you talking about? Force Film is unrelated to 4:2:0/4:2:2.

Re: telecide and y0/y1 usage

Posted: Thu Jun 05, 2014 12:35 pm
by admin
For DGMPGDec, the fieldop is set when making the project in DGIndex. Please refer to the DGIndex user manual.

Video/Field Operation

Perusal of the user manuals is always a valuable pursuit.

Re: telecide and y0/y1 usage

Posted: Thu Jun 05, 2014 5:28 pm
by admin
MrVideo wrote: How tolerant is the software to 3:2 pulldown errors? By this I mean every now and then the equipment used to create the 29.97 tape will drop a frame. For example, the five frame pulldown pattern starts at 0 and ends at 4. Some where in the program, one of the frames is dropped, resulting in the pattern now being frames 4, 0, 1, 2, 3. Is that tolerated by dgindex, or not? telecide/decimate has no problem working with it.
DGIndex just decodes and delivers frames. It is Decomb that needs to adapt to errors, and it does so. If fields are orphaned by losses, they are corrected by the postprocessing, assuming you have it enabled.
Oh, and what percentage of film in the D2V/DGI output is needed before Force Film is viable?
People usually use a standard of 95%, but it really depends on where the non-3:2 stuff is. Usually if you get 95% it's due to interlaced credits at the end, but you have to check that. Formally, postprocessing with FieldDeinterlace() on a Force Film project can fix stuff anywhere, but if the percentage is too low you will get too many frames deinterlaced when they could be field-matched.
EDIT: Nevermind. I just read the tutorial that you pointed to and the material I have doesn't get any higher than 88% film. Most of it is 0%. I'll just stick with telecide/decimate, which provide excellent results.
You can't rely on just this figure. You might have pure interlaced material, which would give 0% but require deinterlacing rather than IVTC. Or it could be hybrid, which presents some dilemmas, such as deciding what output frame rate to target (i.e., whether to decimate). I'm sure you are aware of this, but I mention it for the record.

Re: telecide and y0/y1 usage

Posted: Fri Jun 06, 2014 7:33 am
by admin
MrVideo wrote:I can see where it can come up with 88%. Because the majority of videos that I work with have no pulldown flags, it isn't worth modifying my zshell script to work with so few files, considering that telecide/decimate is doing such a perfect job dealing with all that I throw at it (other than the repeat frame stuff).
A possibility is that there is a combination of hard and soft pulldown. That is really common for anime. In such cases, you have to honor pulldown and IVTC externally, as you are doing.
That is great software. I really like it. It gets constant use.
Great to hear and thank you for your interest in my tools.