XviD Stats Viewer: Analyser for 1st pass log - Page 3

cult · 19th November 2002, 14:46

@Kaiousama
try to run regsvr32 for this dll...

Kaiousama · 19th November 2002, 18:12

@cult: Thanks, now all's going right! (what a lazy am i?

)

@ nytrogen: can you add or even specify that the msstdfmt.dll is required for running the program, on your site? (because i've tested it's not my computer fault the deficency of this dll, because the standard istallation of winXP e win98 doesnt install this file, like the tabctl32.ocx)

Bye

NiTroGen · 19th November 2002, 20:23

@Kaiousama:

The full installation package of version 1.40 was flawed. If you install the full installation package of version 1.20 and then update to v1.51 there is no problem. I've made a new full package of version 1.51, but for some weird reason I cannot upload it to the site. So if anyone needs "msstdfmt.dll", I've attached it below (you can also find it here). Don't forget to "regsvr32" it!

Edit by Swede: I removed the attachment since I'm not really sure it's legit to redistribute it. I could be wrong though and if this is the case I'll put it back.

Tester · 20th November 2002, 21:52

@NiTroGen

hi,

i've testet your program with lord of the rings (special extendet).
3:38h on 3 CD's. The result was very great. Almost perfect image quality and only 180 kb size deviation on 1.9 GB :-)

One question: The "press me" button in the "linear curve scaling" register sets (beside other values...) the bitrate payback delay to 1.
What does this mean? Using the StatsReader of Koepi, he does not mention to set this to 1 (only disabling alt-cc, and set i-frame boost / low/high distance to 0, like yours)

THX

Tester

NiTroGen · 22nd November 2002, 14:30

@Tester:
I'm not an XviD guru (yet!) and I'm still trying to figure what each option does and how it affects the encoding process, so instead of explaining how payback delay works, I'll quote here a small part of Koepi's XviD Options Explained document.

Quote:

The curve compression is controlling the "smoothness" of the scaled bitrate curve. If you set curve compression low/high to 0, you'll disable it, having "perfect VBR" (you have to set payback delay to 1 then as well).
Payback delay is the number of frames which should be used to compensate for under- or overuse of bits

Because we need a "perfect VBR" when we use the linear curve compression, I set payback delay to 1 as well.

Koepi · 22nd November 2002, 14:36

Well, I have to update that info, sorry for the mess!

For perfect VBR, the bitrate achieved for each frame _must_ be the one we predicted. This is very hard to do, and due to some features in xvid simply impossible - therefore this "leaky bucket" overflow system.

Setting payback delay to 1 isn't really closer to perfect VBR as it degrades some frames seriously - it's just a theory made up of an incomplete model (ask dr. google about creating models of the reality and interpreting the results), so reality is: set it to 10-12 seconds, for PAL: 250-300, for FILM 240-264, for NTSC 300-360, it will sure help picture quality.

I hope this helps.

NiTroGen · 22nd November 2002, 16:16

What about the payback method? Should it be with bias to small frames or not, when trying to achieve the best linear downscaling?

Koepi · 22nd November 2002, 17:28

i prefer bias. i think i wrote that in the statsreader thread.

Tester · 22nd November 2002, 22:51

well, I testet 2 encodes with lord of the rings (pal 25 fps) with following settings

disable alt-cc
i-frame bosst: 0
low/high bitrate scenes: 0
quant: 2 - 31 (i, p)
h.263 (both passes)
i-frame: 250, 6

But the first time with bitrate payback 250, and second time with 1 (both bias)

I have the impression that the second one is better and more constant in quality.

final size: 1909000kb
Size deviation of first encode: 186kb
Size deviation of second encode: 1688kb

Is the only disadvantage of setting payback to 1 the size deviation?

@koepi: are these settings usefull also for your newer builds (21.11.2001) and with new functions like qpel and b-frames, or can I use StatsReader, too?

NiTroGen · 27th November 2002, 08:38

OK, new version is up! Now skipped frames can be reported too, and payback delay does not change when you press the "press me" button. And the installation package is different now. VB6's installation system is used, instead of NSIS, and all nessecary DLLs & OCXs are included, except the VB6 runtime libraries of course.

Download the new version here. VB6 runtime DLLs can be found here.

The Link · 27th November 2002, 09:24

I get a runtime error 424 when loading a stats-file! I have installed VB6 runtime before.

Regards,

The Link

NiTroGen · 29th November 2002, 18:03

Quote:

Originally posted by The Link
I get a runtime error 424 when loading a stats-file! I have installed VB6 runtime before.

Regards,

The Link

OK, I've fixed this bug and attached the new executable file. Copy it over the old one.

NiTroGen · 16th February 2003, 17:53

StatsViewer is updated to version 1.60

Quote:

Change log:

+ Added a frame listing window. (Really slow output!!!)
+ Added some presets of XviD settings. Looking forward to your comments, suggestions, etc...
+ Added some simple calculation routines for estimated video size, bits*(pixel/frame), etc...
* Refresh button moved outside the tab panel for easier access.
* Some small bugs fixed.
- Linear curve compression function was removed, because of its implementation in the XviD codec.

As you can see, I have removed the curve scaling function. Now you can find some setting presets, which are automatically set by the press of a button: Linear Curve Scaling (the well-known compression), Normal Curve Scaling for low motion movies (gives a low/high compression of 10%/10%), Alt.Curve Compression (gives a low/high of 100/1000 if no stats file is loaded and the suggested settings if a stats file is loaded).
If you have some presets of your own of some suggestions to make, please post in this topic.

You can find the new version on the Official Site or attached below (please wait to be approved by the moderators).

OBcecado · 16th February 2003, 18:00

Thanks for that new release, Is it safe to use it with b-frames now ?
Keep up the good work.

Best regards.

OBcecado · 16th February 2003, 18:33

I think I've found a bug...
In frame size graph, when the Frame Information button is pressed I get a error message saying : Run-time error '30006': Unable to allocate memory for flexgrid.

NiTroGen · 16th February 2003, 23:45

Quote:

Originally posted by OBcecado
Is it safe to use it with b-frames now ?

Yes it is, because it lets XviD handle the curve and does not alter the stats file.

Quote:

Originally posted by OBcecado
I think I've found a bug...
In frame size graph, when the Frame Information button is pressed I get a error message saying : Run-time error '30006': Unable to allocate memory for flexgrid.

I haven't tested with a "large" stats file. My stats contained about 20000 frames. I'll finish an encode of a full movie tomorrow (about 190000 frames) to check your report.

ookzDVD · 18th February 2003, 08:13

@Nitrogen,

The latest v1.6 is awesome!

azra28 · 18th February 2003, 10:16

Is someone could get me an adress because I can't acces to site

cult · 18th February 2003, 17:55

there u go:
http://homepages.pathfinder.gr/statsviewer

azra28 · 18th February 2003, 19:51

Thks to u

19th November 2002, 18:12	#42 \| Link
Kaiousama Italian Doom9 Staff Join Date: Apr 2002 Location: Italy Posts: 219	@cult: Thanks, now all's going right! (what a lazy am i? ) @ nytrogen: can you add or even specify that the msstdfmt.dll is required for running the program, on your site? (because i've tested it's not my computer fault the deficency of this dll, because the standard istallation of winXP e win98 doesnt install this file, like the tabctl32.ocx) Bye __________________ "All that we see or seem is but a dream within a dream" E.A.Poe Last edited by Kaiousama; 19th November 2002 at 18:15.

19th November 2002, 20:23	#43 \| Link
NiTroGen aka Íéêüëáïò Join Date: Mar 2002 Location: Ioannina, Greece Posts: 125	@Kaiousama: The full installation package of version 1.40 was flawed. If you install the full installation package of version 1.20 and then update to v1.51 there is no problem. I've made a new full package of version 1.51, but for some weird reason I cannot upload it to the site. So if anyone needs "msstdfmt.dll", I've attached it below (you can also find it here). Don't forget to "regsvr32" it! Edit by Swede: I removed the attachment since I'm not really sure it's legit to redistribute it. I could be wrong though and if this is the case I'll put it back. __________________ NiTroGen: Colorless, odorless, tasteless, generally inert gas. Fifth most abundant element in the universe. Makes up about 78% of earth's atmosphere.

22nd November 2002, 14:36	#46 \| Link
Koepi Moderator Join Date: Oct 2001 Location: Germany Posts: 4,454	Well, I have to update that info, sorry for the mess! For perfect VBR, the bitrate achieved for each frame _must_ be the one we predicted. This is very hard to do, and due to some features in xvid simply impossible - therefore this "leaky bucket" overflow system. Setting payback delay to 1 isn't really closer to perfect VBR as it degrades some frames seriously - it's just a theory made up of an incomplete model (ask dr. google about creating models of the reality and interpreting the results), so reality is: set it to 10-12 seconds, for PAL: 250-300, for FILM 240-264, for NTSC 300-360, it will sure help picture quality. I hope this helps. __________________ Koepi's new media development site

22nd November 2002, 16:16	#47 \| Link
NiTroGen aka Íéêüëáïò Join Date: Mar 2002 Location: Ioannina, Greece Posts: 125	What about the payback method? Should it be with bias to small frames or not, when trying to achieve the best linear downscaling? __________________ NiTroGen: Colorless, odorless, tasteless, generally inert gas. Fifth most abundant element in the universe. Makes up about 78% of earth's atmosphere.

22nd November 2002, 17:28	#48 \| Link
Koepi Moderator Join Date: Oct 2001 Location: Germany Posts: 4,454	i prefer bias. i think i wrote that in the statsreader thread. __________________ Koepi's new media development site

19th November 2002, 14:46	#41 \| Link
cult Registered User Join Date: Nov 2001 Location: greece Posts: 383	@Kaiousama try to run regsvr32 for this dll...

20th November 2002, 21:52	#44 \| Link
Tester Member Join Date: Apr 2002 Location: Germany - NRW Posts: 26	@NiTroGen hi, i've testet your program with lord of the rings (special extendet). 3:38h on 3 CD's. The result was very great. Almost perfect image quality and only 180 kb size deviation on 1.9 GB :-) One question: The "press me" button in the "linear curve scaling" register sets (beside other values...) the bitrate payback delay to 1. What does this mean? Using the StatsReader of Koepi, he does not mention to set this to 1 (only disabling alt-cc, and set i-frame boost / low/high distance to 0, like yours) THX Tester

22nd November 2002, 22:51	#49 \| Link
Tester Member Join Date: Apr 2002 Location: Germany - NRW Posts: 26	well, I testet 2 encodes with lord of the rings (pal 25 fps) with following settings disable alt-cc i-frame bosst: 0 low/high bitrate scenes: 0 quant: 2 - 31 (i, p) h.263 (both passes) i-frame: 250, 6 But the first time with bitrate payback 250, and second time with 1 (both bias) I have the impression that the second one is better and more constant in quality. final size: 1909000kb Size deviation of first encode: 186kb Size deviation of second encode: 1688kb Is the only disadvantage of setting payback to 1 the size deviation? @koepi: are these settings usefull also for your newer builds (21.11.2001) and with new functions like qpel and b-frames, or can I use StatsReader, too?

27th November 2002, 08:38	#50 \| Link
NiTroGen aka Íéêüëáïò Join Date: Mar 2002 Location: Ioannina, Greece Posts: 125	New Version OK, new version is up! Now skipped frames can be reported too, and payback delay does not change when you press the "press me" button. And the installation package is different now. VB6's installation system is used, instead of NSIS, and all nessecary DLLs & OCXs are included, except the VB6 runtime libraries of course. Download the new version here. VB6 runtime DLLs can be found here. __________________ NiTroGen: Colorless, odorless, tasteless, generally inert gas. Fifth most abundant element in the universe. Makes up about 78% of earth's atmosphere.

27th November 2002, 09:24	#51 \| Link
The Link Registered User Join Date: Nov 2001 Location: Germany Posts: 189	Bug? I get a runtime error 424 when loading a stats-file! I have installed VB6 runtime before. Regards, The Link __________________ When birds burp, it must taste like bugs! (Calvin&Hobbes)

16th February 2003, 18:00	#54 \| Link
OBcecado LuLz Join Date: Oct 2002 Location: Portugal Posts: 113	Thanks for that new release, Is it safe to use it with b-frames now ? Keep up the good work. Best regards.

16th February 2003, 18:33	#55 \| Link
OBcecado LuLz Join Date: Oct 2002 Location: Portugal Posts: 113	I think I've found a bug... In frame size graph, when the Frame Information button is pressed I get a error message saying : Run-time error '30006': Unable to allocate memory for flexgrid.

18th February 2003, 08:13	#57 \| Link
ookzDVD DVD Rebuilder! Join Date: Oct 2001 Posts: 1,147	@Nitrogen, The latest v1.6 is awesome!

18th February 2003, 10:16	#58 \| Link
azra28 Registered User Join Date: Mar 2002 Location: France Posts: 9	Is someone could get me an adress because I can't acces to site

18th February 2003, 17:55	#59 \| Link
cult Registered User Join Date: Nov 2001 Location: greece Posts: 383	there u go: http://homepages.pathfinder.gr/statsviewer

18th February 2003, 19:51	#60 \| Link
azra28 Registered User Join Date: Mar 2002 Location: France Posts: 9	Thks to u

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode