Doom9's Forum - View Single Post

manono · 2nd November 2003, 02:06

Hi-

I've done a couple of movies now using gkItFast4U. Did them both for 1 CD. The speeds seem about the same as before. Perhaps those finding it to be slower are used to using different codecs, or have been using XviD with different settings. If, for example, you haven't been using VHQ, since gkItFast4U does use VHQ, then you'll get a pretty big slowdown. And if you're not used to using LanczosResize, then that slows you down a little bit.

The first movie had 82% Film. It detected that, put on IVTC, ran the compression test and encoded. I preset the Horizontal Res to 640, so it kept that and came out with a 74% compress test result. It made the subs, did the audio, cropped the black bars perfectly, encoded the movie and muxed with no problem, and it looks great. Perfect 700 MB size.

The second movie it handled differently. I had done both movies before, so I knew what I was looking for. This time I changed the final file size to 701 MB, from the 700 MB of the first movie. It kept the AC3 track. It looks like anything over 700 MB gets AC3 (I just confirmed that in the Usage.txt). Many people might raise the size slightly, still intending to make 1 CD (perhaps even overburning). One suggestion is to make MP3 audio up to about 705 MB. Either that or let us choose the Audio in the Advanced Settings box. Some people might prefer MP3 Audio, even for a 2 CD rip. This time I let it choose the Horizontal Resolution (Auto Width). This one had 0% Film in DVD2AVI, but I knew from experience that it IVTC'd well. It kept it at 29.97fps, kept the AC3 audio, did the subs, chose a softer matrix, cropped properly, kept LanczosResize interestingly, and chose a resolution based on a 74% compress test result. The resolution wound up as 480x256. I don't know any way around deciding to IVTC or just Deinterlace based just on the DVD2AVI results. You have to have a look at it usually. I don't know if you want to add an IVTC box in the Advanced settings or not. Heck, just deciding to Deinterlace or not follow the ITU Standard is going to confuse a lot of beginners. In fact, I'd suggest not giving anyone that particular choice. Make the ITU Standard the default and don't give anyone the chance to change it. But it came out to exactly 701 MB, and except for looking a bit soft, turned out fine.

A lot of these issues that will crop up are judgement calls. And you've made some reasonable choices in order to keep it as simple as possible. And it's real simple. A person that has never encoded anything in his life will have no problem at all with this. This is as close to a click-and-forget program as I've ever seen. And that includes Fair Use.

Lets see-questions.
1. Is there any point in running a compress test if the user has already chosen a Horizontal Resolution? Perhaps to show why his movie didn't turn out as well as expected after it's finished? If someone comes around later and says his movie looks like crap, you can say, "Post the log.", and then tell the guy, "Hey, you got a 30% compress test result with 640 Horizontal Res for 1 CD. What did you expect?"

2. What Deinterlacer are you using, and at what settings? Maybe I'll do that second movie again with the deinterlacer turned on and find out.

3. Which Matrices are you using? When it said "Using Softer Matrix", did it go from HVS_Best to HVS_Good, by chance?

4. What does the "Normal" in "4:3 Normal" mean? It's in Video Info, below the Input Directory. It was a widescreen letterboxed 4:3 DAR movie.

5. It looks like you choose the Horizontal Resolution based on around 75% in the compress test results. I'd suggest that if you're in danger of going below, say 512*xxx, that you switch to a 60 or 65% result, and perhaps also switch to Bicubic resizing. That may be tricky to implement, so I'd understand if you don't much like the idea. And I realize that you can choose a minimum Hor. Res. in the Advanced Settings.

6. Maybe add a line in the log with the actual final percentage based on the actual first pass size. When you're running a compress test at 640*xxx, and then extrapolating to get the percentage for a different resolution, that, on top of the compress test's margin of error, may give up to, what, 10% or so difference?

I guess that's it for now. My suggestions are minor, and even if you change nothing, this thing's great. Now when people complain that GKnot's too hard to use, I can say, "Use gkItFast4U." I'm impressed. Good work, len0x.

2nd November 2003, 02:06	#18 \| Link
manono Moderator Join Date: Oct 2001 Location: Hawaii Posts: 7,406	Hi- I've done a couple of movies now using gkItFast4U. Did them both for 1 CD. The speeds seem about the same as before. Perhaps those finding it to be slower are used to using different codecs, or have been using XviD with different settings. If, for example, you haven't been using VHQ, since gkItFast4U does use VHQ, then you'll get a pretty big slowdown. And if you're not used to using LanczosResize, then that slows you down a little bit. The first movie had 82% Film. It detected that, put on IVTC, ran the compression test and encoded. I preset the Horizontal Res to 640, so it kept that and came out with a 74% compress test result. It made the subs, did the audio, cropped the black bars perfectly, encoded the movie and muxed with no problem, and it looks great. Perfect 700 MB size. The second movie it handled differently. I had done both movies before, so I knew what I was looking for. This time I changed the final file size to 701 MB, from the 700 MB of the first movie. It kept the AC3 track. It looks like anything over 700 MB gets AC3 (I just confirmed that in the Usage.txt). Many people might raise the size slightly, still intending to make 1 CD (perhaps even overburning). One suggestion is to make MP3 audio up to about 705 MB. Either that or let us choose the Audio in the Advanced Settings box. Some people might prefer MP3 Audio, even for a 2 CD rip. This time I let it choose the Horizontal Resolution (Auto Width). This one had 0% Film in DVD2AVI, but I knew from experience that it IVTC'd well. It kept it at 29.97fps, kept the AC3 audio, did the subs, chose a softer matrix, cropped properly, kept LanczosResize interestingly, and chose a resolution based on a 74% compress test result. The resolution wound up as 480x256. I don't know any way around deciding to IVTC or just Deinterlace based just on the DVD2AVI results. You have to have a look at it usually. I don't know if you want to add an IVTC box in the Advanced settings or not. Heck, just deciding to Deinterlace or not follow the ITU Standard is going to confuse a lot of beginners. In fact, I'd suggest not giving anyone that particular choice. Make the ITU Standard the default and don't give anyone the chance to change it. But it came out to exactly 701 MB, and except for looking a bit soft, turned out fine. A lot of these issues that will crop up are judgement calls. And you've made some reasonable choices in order to keep it as simple as possible. And it's real simple. A person that has never encoded anything in his life will have no problem at all with this. This is as close to a click-and-forget program as I've ever seen. And that includes Fair Use. Lets see-questions. 1. Is there any point in running a compress test if the user has already chosen a Horizontal Resolution? Perhaps to show why his movie didn't turn out as well as expected after it's finished? If someone comes around later and says his movie looks like crap, you can say, "Post the log.", and then tell the guy, "Hey, you got a 30% compress test result with 640 Horizontal Res for 1 CD. What did you expect?" 2. What Deinterlacer are you using, and at what settings? Maybe I'll do that second movie again with the deinterlacer turned on and find out. 3. Which Matrices are you using? When it said "Using Softer Matrix", did it go from HVS_Best to HVS_Good, by chance? 4. What does the "Normal" in "4:3 Normal" mean? It's in Video Info, below the Input Directory. It was a widescreen letterboxed 4:3 DAR movie. 5. It looks like you choose the Horizontal Resolution based on around 75% in the compress test results. I'd suggest that if you're in danger of going below, say 512xxx, that you switch to a 60 or 65% result, and perhaps also switch to Bicubic resizing. That may be tricky to implement, so I'd understand if you don't much like the idea. And I realize that you can choose a minimum Hor. Res. in the Advanced Settings. 6. Maybe add a line in the log with the actual final percentage based on the actual first pass size. When you're running a compress test at 640xxx, and then extrapolating to get the percentage for a different resolution, that, on top of the compress test's margin of error, may give up to, what, 10% or so difference? I guess that's it for now. My suggestions are minor, and even if you change nothing, this thing's great. Now when people complain that GKnot's too hard to use, I can say, "Use gkItFast4U." I'm impressed. Good work, len0x. Last edited by manono; 3rd November 2003 at 05:38.