Decomb 1.91 functioning very slowly?!

[samfchen]

Hello,

I tried Decomb 1.8X and that was functioning rather well under Gknot, around 10fps+ on my mobile P3-1 ghz. But when I switched over to Decomb 1.91 (newest version) on another system (AMD XP 2000+), encoding speed seemed to be VERY slow, comparatively speaking. I was only getting around 14fps (in ANY reasonable situation pretty much, the AMD machine should encode at 20+fps). I haven't yet gotten a chance to try Decomb 1.91 on the mobile P3, but I imagine it will be much slower then 10fps. If that is in fact the case, is there something "wrong" with Decomb 1.91; and if that's not the case, then is there something wrong with Decomb 1.91 AND AMD?


Thanx in advance,
Sam:)

[manono]

Hi samfchen-
I'm not positive, but pretty sure this is the reason:

Using v 1.81, you had to have the extra line of FieldDeinterlace if you wanted to deinterlace. I'm guessing you didn't put in that line.

Using v 1.91 FieldDeinterlace is now default in Telecide, and that slows things considerably. If you want to disable it, you have to add some stuff after Telecide-it's in the Help file how to do it.

[Dali Lama]

Do you have to use Field Deinterlace with anime?

Like Kenshin?

You seem to know a lot about anime encoding.

Thanks,

Dali

[western shinma]

That depends on the anime. For Kenshin, I think you would need it, although some shows are fine without any additional deinterlacing. If you don't want to be surprised later you might want to just leave it on all the time, since the default settings have been found to work well on most material.

[Dali Lama]

Thanks makes sense shinma. It is better to be safe than sorry. Also, thanks for the input about Kenshin.

Dali

[Guest]

Yes, the usual reason people think 1.91 is slow is because they are comparing its default postprocessing to the previous Telecide without FieldDeinterlace, as manono observed.

There is another important speed tip. If you have Telecide or Decimate without any parameters, like this:

Telecide
Decimate

replace it with these:

Telecide()
Decimate()

That will give you a 25-33% speed increase! I don't know why for sure yet, but many people have confirmed it. Probably C++ virtual function craziness.