Username:
B
I
U
S
"
url
img
#
code
sup
sub
font
size
color
smiley
embarassed
thumbsup
happy
Huh?
Angry
Roll Eyes
Undecided
Lips Sealed
Kiss
Cry
Grin
Wink
Tongue
Shocked
Cheesy
Smiley
Sad
<- 1  -   of 69 ->
--
--
List results:
Search options:
Use \ before commas in usernames
Edit history:
bmn: 2010-10-14 01:27:11 am
bmn: 2010-10-14 01:25:27 am
bmn: 2010-10-14 01:23:50 am
bmn: 2010-10-14 01:23:28 am
bmn: 2010-10-14 01:22:15 am
bmn: 2010-10-14 12:55:56 am
bmn: 2010-10-14 12:52:57 am
bmn: 2010-10-14 12:50:58 am
bmn: 2010-10-14 12:43:58 am
bmn: 2010-10-14 12:29:34 am
bmn: 2010-10-14 12:29:12 am
bmn: 2010-10-14 12:28:19 am
It's F2 because the game itself only runs at ~30fps. So encoding a video at F1 for an F2 game is just a waste of resources/bitrate/etc as each set of two frames are basically identical. It's basically the same as how runs of older consoles (which have a native res of D4) are encoded at D4 at all qualities despite the recording being D1.

Anri cropping: it doesn't have that functionality right now. Once you've told anri what source video files to use, it'll make a file called [project]_source.avs in the project directory. This file loads the videos so that the individual qualities can work on them, so anything you do at the end of that file will affect the encoded videos.

The command you want is Crop(0, 3, -3, -0) (see Avisynth/Crop). Just put it at the end of the file, e.g.:

Code:
avisource("D:\Source Video\MGS2S - Missions\raiden\01-01.avi")
converttorgb32
crop(0, 3, -3, -0)


This method only works in anri 3.3 (the version being worked on right now, which bos has given you). 3.2's Avisynth script layout is a lot different.

Edit: Actually I just took a look at the video you posted, and it looks more like it's replaced the top and right 3 pixels with black than actually added them. The video should be 1280x720 with no border, but it's 1280x720 with one. My capture card does the opposite, and produces 720x486 video instead of 720x480.

If you do the cropping, that'll put it down to 1277x717, which will mess with the aspect ratio. Anri always resizes the video to the AR you choose, but you want to avoid an actual resize if possible, so it may just be best to leave it as is. If you really want to get rid of the border though, then feel free, it won't hurt too badly.
Edit history:
TheThrillness: 2010-10-14 01:31:02 am
thethrillness.blogspot.com
Is there a way to actually tell what the game runs at? I ask this because obviously that capture was 59.94 but really 30fps as you said. If I capture a game that is actually a full 60fps like Street Fighter IV, how can I detect what is a true 60 and what is 30?

Thanks. Will encode with and without and see what looks better.
You basically do it by playing the video and looking at each frame individually. In VLC for example, you can go forward one frame by pressing E.

When the game is F1, every frame will be different (assuming no slowdown etc). If it's F2, then you'll have to advance two frames before the picture changes.
thethrillness.blogspot.com
Quote from bmn:
You basically do it by playing the video and looking at each frame individually. In VLC for example, you can go forward one frame by pressing E.

When the game is F1, every frame will be different (assuming no slowdown etc). If it's F2, then you'll have to advance two frames before the picture changes.


Ah. I wondered why my player was doing that. Nice little trick. Thanks again. Now to get this run done.
Bug in the statID. Worked fine in 3.2.

Attachment:
Edit history:
bmn: 2010-10-17 12:08:16 pm
I believe what's happening is the first ) is wrongly being treated as the closing bracket for the echo block in :saveavssettings. If that's the case and it can't be avoided then we may have to abandon that syntax.
Edit history:
ballofsnow: 2010-10-23 11:58:46 am
Yay batch code! Does it screw up on quotations as well? -yep

Currently the quotations get replaced by chr(34), the parentheses may screw it up.

SET statid1=%statid1:"="+chr(34)+"%


Unrelated idea: is there a way to tell if two avisynth scripts produce the exact same output? A comparison of the text wouldn't work, i.e. quality=mq in one, and hq in the other. If we can find out whether they are the same, we may save on the amount of x264 passes that need to be performed. Imagine doing a first pass only once for whichever qualities, and second pass for each to get their respective bitrate. Something to think about...
Edit history:
bmn: 2010-10-21 05:19:38 pm
bmn: 2010-10-21 04:07:46 pm
bmn: 2010-10-21 04:05:33 pm
bmn: 2010-10-21 04:04:51 pm
bmn: 2010-10-21 03:41:41 pm
bmn: 2010-10-21 03:38:56 pm
bmn: 2010-10-21 03:26:50 pm
What actually goes into the logfile on a first pass? Is it just how many bits would be required for full quality at each part of the video, and not affected by the stated bitrate? I like this idea if so.

Because the content is identical, anything that can return the dimensions and frame rate should be suitable. avsinfo prints out a load of stats including those two. Or we could use any AVS-capable app to open each of the qualities with some WriteFile code to write it to a file, and then compare in anri...

The same idea could be used (...unless it already is?) for the audio, except that'd be easier for obvious reasons.

Edit: There's also avs2wav in that package?

Code:
# avsinfo.exe -m --width --height --fps "C:\Users\Ian\Desktop\a3-test\a3-test_XQ.avs"
600
448
59.9401
thethrillness.blogspot.com
Not really related to anri chan but I was wondering if you could help me. I now need to time my main game run before I can slap a finish time on each chapter I encode with anri-chan.

As a tester I used Chapter 1. Manually timing it with frames is a total of 31,369. Sony Vegas reads this as 8:43:10 while an online converter reads 8:43:20. Original framerate is 59.940. I typed that into the online calculator too. Is Vegas correct?
Manual calculation gives 00:08:43:340.
So I suppose it depends on the amount of precision they use.
Edit history:
bmn: 2010-11-05 03:06:57 pm
Do fractions of a second matter in the game you're running?

SDA manual timing's done using frame counts IIRC so I'd suggest just using a calculator to work it out:

frame_count / 60 * 1.001
1-Up!
What is the significance of the *1.001?
The framerate isn't 60 FPS. It's 59.940.
fps / 60 / 1.001 = fps / (60 * 1.001) = fps / 59.940
Edit history:
TheThrillness: 2010-11-05 03:34:01 pm
thethrillness.blogspot.com
Quote from bmn:
Do fractions of a second matter in the game you're running?


It's Heavy Rain so not really. I just wanna really know a sure fire way to do it. For example if my run was registered as 8:43:26 on Vegas then 8:43:36 on another application, what I can trust so I can round down (or up) the second.
1-Up!
Quote from Mystery:
The framerate isn't 60 FPS. It's 59.940.
fps / 60 / 1.001 = fps / (60 * 1.001) = fps / 59.940

Guess I'm still confused.  I suspected that the *1.001 was intended to account for the fact that games dont run at 60fps perfectly, but
60 * 1.001 isn't 59.940, it's greater than 60 (because 1.001 > 1)
Edit history:
Mystery: 2010-11-05 03:47:08 pm
I'm guessing that it's a typo (left out paranthesises).
That is, it should be
frame_count / (60 * 1.001)
Edit history:
TheThrillness: 2010-11-05 04:10:39 pm
thethrillness.blogspot.com
If you read above I was told that the source file is actually 29.970 since in a media player I need to press twice to advance 1 frame. I changed Vegas to this frame rate and timed. 15,687 now. Now both Vegas and the online calculator agree that this is 8:43:13
What? The number of frames is constant. How many frames are there really?
The fps merely changes the playtime.
thethrillness.blogspot.com
Sorry I don't understand. I'll try and explain in greater detail. Although the capture card captures 59.94, all of the game is 29.97 so I changed Vegas to preview the project at 29.97 for new frame values. I will also be encoding it at F2 in anri chan.
@Flip, Mystery:

NTSC's frame rate is exactly 60 / 1.001, which is slightly higher than 59.94. The multiplication confusion comes from when I said use a calculator, which means that parentheses aren't available on a basic calculator. The equation I posted is correct, if confusing because NTSC seems to be 60.06 instead, but with parentheses it would be f / (60 / 1.001).
1-Up!
That makes a lot more sense.  Thanks for clarifying!
Edit history:
ballofsnow: 2010-11-07 12:18:40 pm
http://forum.speeddemosarchive.com/post/batman_arkham_asylum_quality_test.html

AviSynth is a bottleneck, looking into it.

Simple test: take LQ avs script into vdub and produce a new master file, I used Lagarith codec in this case. Edit LQ avs in notepad so it only has the avisource line to your new master file. Run job.bat to get back into the encoding part, encode LQ and watch it fly...

My test was from a 1920x1200@30p source, so it's probably lanczos4resize...

^ oh yeah, comparing GUI screens to command line sucks.. any way you can get the command line parameters from avidemux? Probably doesn't matter anyway since AviSynth seems to be the problem.

edit-
BilinearResize gave a speed increase, but not maxing out the CPU.
Could it be an IO bottleneck? Doubtful since the data rate for my file is only 50 MB per sec and it's on a SSD.

May need to look into MT.
http://forum.doom9.org/showthread.php?p=1412274#post1412274

edit-
AviSynth MT is an improvement, but still not maxing cpu.

What do you guys think about encoding all the qualities...at the same time?

I've got the above example doing all L/M/H/I/XQ at the same time maxing out 8 logical cores.
Edit history:
TheThrillness: 2010-11-07 12:44:03 pm
TheThrillness: 2010-11-07 12:36:38 pm
TheThrillness: 2010-11-07 12:34:40 pm
thethrillness.blogspot.com
I was actually just about to post about CPU usage. I am encoding my first chapter and checked CPU usage randomly. The LQ I remember was 40% CPU usage. Now I am encoding the HQ version and CPU usage at 90% (peaks at about 98).

http://img221.imageshack.us/img221/9840/anri1.png

My CPU is only an i3 but it should be able to encode faster than 3 fps at HQ (LQ was 9.5 fps IIRC)

What annoys me is that I can input the same m2ts file into Vegas at 720p, stick the bitrate to 10,000 in Sony's AVC codec and get near enough pixel perfect picture with the source in a little under 7 minutes of encoding instead of 2+ hours on anri chan.

Edit - My HQ encode will not go over 3.17 fps. Definately AviSynth.

Also turned off anti virus, now at a full 100% of CPU.

http://img35.imageshack.us/img35/3083/anri2.png
I just took a look at the encoding settings for Rosenkreuzstilette, which was published not too long ago. It turns out that it used ref=3 when encoding.
If this was encoded with anri, then seriously? That is an awful small amount. Unless device compatibility is desired, this should be increased as much as possible. The very slow preset uses 16.
^ Why post this? You already know 3.3 will use the very slow preset.