EclPRO v0.52 - ProCoder support in DVD-RB (now v1.10)

[robot1]

Thanks again to borgraf for his extensive testing.
I've solved the little setup bug...
Anyway, I'm glad we have solved the Error 5/Stuttering problem: I hope now compatibility with DVD-RB is almost perfect.

[TheSeeker]

Ok Im getting an unusual error I haven't seen with the Procoder/DVD-RB combo before. I have attached an image file of the error that I see.. So I guess I have to wait mod approval but here is my setup. Im using avisynth 2.56 alpha, dvd rb .74, dgdecode 1.13b, procoder 2, eclpro .25a. I have everything pointed to the correct files in rebuilder setup and i have successfully used eclpro .24 before, but not .25a. I have the strictgopbratecontrol=1 and alwaysgop15=0 options set. I run prepare on the source and all is fine. I go to encode and procoder opens fine and then it seems to run pretty fast through an encode (1.20 of realtime, which is faster than I have ever seen it go), then it closes and errors with the attached error code. Then I hit ok on the error window and it opens the next file and seems to do something then that closes and erros when it gets to the end and it keeps doing this until I abort encode on dvd rb. So just wondering what could be causing this? Any ideas anyone?

[borgraf]

I'm now on 3 successful rebuilds with ECLPro 0.25a. I noticed you have a whole lot of alpha and beta software in that setup. It's a longshot but you could try the same setup as me on the off chance that it matters. I'm using AviSynth 2.5.5, DGDecode from DGMPGDec 1.1.0 Final, AutoParameters=1, AlwaysGOP15=0, StrictGOPBrateControl=0

just a suggestion, good luck

[robot1]

Originally posted by TheSeeker
So just wondering what could be causing this? Any ideas anyone? Just to point out two things:
1 - EclPRO needs rmseq.exe, and looks for it in the same directory (I will add a check for next version that warns if the program is missing)
2 - from our PM's we have established that rmseq.exe doesn't work with WindowsXP 64 beta. If you have installed this beta version of the S.O. and want to use Procoder, the last working version is EclPRO v0.22. Anyawy, looking to solve this issue.

[wmansir]

I think I have discovered an problem with EclPro's handling of interlaced content. Some testing I have done with NuMenu4U appears to show that EclPro is not setting the TopFieldFirst flag correctly. Of course this causes interlaced output to stutter.

I'm not sure if this parameter was overlooked because DVD-RB does not make use of it, or perhaps it is reversed. I haven't done extensive testing yet. One file I did try was a CCE 2.5 .ecl that had top_first=0, CCE output an MPEG with TFF=1. The same .ecl done with Procoder had TFF=0.

Later versions of CCE use the offset_line= parameter instead of (or in addition to) top_first in order to set field order. I haven't tested them.

CCE has always had a messed up way of handling TFF settings, so the .ecl setting may not match the final output. I think the important thing here is that EclPro match CCE's output as closely as possible.

This issue doesn't affect DVD-RB because the TFF flag is overwritten by DVD-RB in the rebuild phase, so any error is corrected in the final output.

[robot1]

Originally posted by wmansir
I think I have discovered an problem with EclPro's handling of interlaced content. Some testing I have done with NuMenu4U appears to show that EclPro is not setting the TopFieldFirst flag correctly. Of course this causes interlaced output to stutter.

I'm not sure if this parameter was overlooked because DVD-RB does not make use of it, or perhaps it is reversed. I haven't done extensive testing yet. One file I did try was a CCE 2.5 .ecl that had top_first=0, CCE output an MPEG with TFF=1. The same .ecl done with Procoder had TFF=0.
Later I will test and fix. I don't know if I've implemented both the CEC 2.5 and 2.6x parameters. DVD-RB uses the 2.5 one, while NuMenu4u the 2.6x.


CCE has always had a messed up way of handling TFF settings, so the .ecl setting may not match the final output. I think the important thing here is that EclPro match CCE's output as closely as possible.
Of course!



This issue does effect DVD-RB because the TFF flag is overwritten by DVD-RB in the rebuild phase, so any error is corrected in the final output. [/B] Does the issue affect DVD-RB?
From what I've understood, it doesn't :confused:
Anyway, it will fixed.

[wmansir]

Whoops, I meant doesn't affect DVD-RB. I fixed it.

Also, now that I'm recalling it, CCE 2.5 (and maybe all versions) output TFF regardless of the Top_first= setting. That setting's true effect in 2.5 is to shift the input video 1 line (like offset_line=1), effectively making BFF input into TFF video and visa versa.

So the Top_first= setting is actually reversed.

Top_First=0

CCE outputs TFF, input is left untouched.

ProCoder should be set to TFF.


Top_First=1

CCE outputs TFF, Input shifted 1 line

ProCoder should output BFF.

The Offset_Line parameter made this process more apparent, but caused more confusion than anything.

[robot1]

This should solve the bff/tff issue reported with NuMenu4U.

[EDIT]
It's not perfect yet, because NuMenu4U always outputs
progressive=1
even for interlaced sources.

So, don't use ProCoder with interlaced menus till next NuMenu4U version.

[SlipGun]

I was just wondering, what are the AutoParameters bitrate limits for interlaced sources? Also, is DC Prec something you can set in eclpro.ini like GOP length?

[robot1]

for interlaced sources, bitrates limits are raised by 150.
If autoparameters=0, EclPRO uses DCPrec defined in the .ecl

[wmansir]

I've encountered a special case where EclPro doesn't output the same as CCE and causes a frame count mismatch errror in DVD-RB's rebuild phase.

This happens when DVD-RB builds a 0 frame job. Whether RB should be building 0 frames jobs is another matter. I think it's due to VobBlanker, since I've seen this before when using it.

CCE creates a 0 byte file, where EclPro creates a 1 frame video.

I've attached a file with all the relevent files. I even included the rebuilder.inf and VTS .ifo in case jdobbs or anyone else is interested. Obviously the .vob is too large to include but I pasted the vobedit info from the first video and nav header into a text file, since this 0 frame segment is the first one.

[robot1]

I could modify EclPRO to check the number of frames, and if is=0, to create a 0 byte file, and not run ProCoder.
Do you think this would solve?

[wmansir]

That sounds fine. I think you'll want to test for "encode_last=0" in the ecl. Or "last_frame=0", though going by encode_last will emulate CCE's behavior better.

[robot1]

Changelog:
Changed behaviour for files with no frames to encode, to reflect CCE's one.

Thanks to wmansir for discovering the bug, and for the suggestions.

The download will be available as soon as a mod approves it.

[jdobbs]

Originally posted by wmansir
I've encountered a special case where EclPro doesn't output the same as CCE and causes a frame count mismatch errror in DVD-RB's rebuild phase.

This happens when DVD-RB builds a 0 frame job. Whether RB should be building 0 frames jobs is another matter. I think it's due to VobBlanker, since I've seen this before when using it.

CCE creates a 0 byte file, where EclPro creates a 1 frame video.

I've attached a file with all the relevent files. I even included the rebuilder.inf and VTS .ifo in case jdobbs or anyone else is interested. Obviously the .vob is too large to include but I pasted the vobedit info from the first video and nav header into a text file, since this 0 frame segment is the first one. I don't see the "0 frame job" in your post anywhere??? I do see a single frame job...

[EDIT] I do see, though the DVD-RB is putting frames=0 in the INF file for a job that has trim(0,-1) in it. That trim setting actually returns one cell (frame 0).

Has any of the files in your zip been manually modified? I just looked at the code, and the two values are output within a few lines of each other... I don't see how a job that has "trim(0,-1)" in it could output a frame count of zero???

[wmansir]

The .avs is 1 frame, but the .ecl has last_frame=0 and encode_last=0, which is why nothing is encoded.

[jdobbs]

Could you do me a favor and tell me what the trim() values are for the next segment following that one? I think I may have it.

[wmansir]

I don't have the original files anymore, I could try and recreate them, but I did some pre-processing so I can't guarantee I can recreate the problem.

I do know the next segment had 1 frame output, but I don't recall the trim values.

[wmansir]

I should add that when I previously encountered this situation the trim values in the next segment were the same as the first (0,-1). But I don't recall well enough to say so in this case. And in both instances that I can recall the 0-frame segment was the first in the VTS.

[jdobbs]

That's ok -- I think I know how it happened. Thanks...

[robot1]

@Jdobbs
ProCoder encodes always inserting the sequence endcode for every segment.
When it's not needed EclPRO runs another program (rmseq.exe) which scans the output and remove the sequence.
As DVD-RB now handles correctly files with sequence endcode (even if it's not needed), could you use in the .ecl always the sequence endcode "on" (so EclPRO doesn't call the external program)?
This could lessen the total encoding time using ProCoder.

[jdobbs]

Ok... just so I have it right. You want me to always turn sequence endcode on in the .ECL for Procoder encodes... that should be easy enough.

[jdobbs]

Done. It'll be in v0.86

[robot1]

Thank you.
:)

[daphy]

Updates:
- Added support for Canopus HQ format output.
- Fixed issue when outputing files larger than 2GB.
- Fixed issue when converting to Windows Media files.
- Fixed issue when applying multiple audio filters.
- Fixed issue with field order in exported file.
- pulldown error fixed for dvd profile

ftp://ftp.canopus.co.jp/pub/jp/drivers/win2k/

[Blackout]

hey guys, im getting a problem with Procoder 2.0 and Rebuilder where there are slight skips in the films that are encoded, when there are chapter points. Sometimes the audio goes slightly out of sync, until the next chapter point. Im guessing that something is passing too many or incorrect frames at the chapter points to Procoder. Or maybe its ECLPro, i was using ver 1.21 at the time of the encode. It was a hollywood dvd.

With CCE doing the encoding, everything works fine and there are no skips, so its either ECLPro or PRocoder 2.0...anyone seen this problem and had it sorted?

many thanks,
Blackout

[robot1]

Probably it was the sequence endcode inserted at the end of every cell.
It was fixed in EclPRO 0.25
Now DVD-RB can handle .m2v with the sequence endcode, but I don't know if a simple "No Compression" mode will fix: you could try on a RW.
If it doesn't fix, you have to repeat the whole process.

[D3s7]

Finally getting around to playing w/ procoder and eclpro and I'm running into the "Target File Warning" box (2.04.02)

I've up'd the delay setting in the .ini but doesn't seem to have any impact..

Anyone w/ any additional thoughts?

[robot1]

I can't repeat the error - probably my PC is much slower :(
If you could PM me a valid email address, I could send you a test version tomorrow.

[Blackout]

Hi, ive gone back to a new re-encode from new VIDEO_TS folder files, this time using EClPro v0.27 with Rebuilder 0.93 and Procoder V2.01. The new encode still has the slight stutter/skips at index changes. The original naturally does not have this (Blackadder II) and an encode with DVDShrink does not have this problem either (although the quality is not as good...which is why i really want DVD-RB and PRocoder to work!!).

I wander if the fact that its a PAL DVD, and that its from Interlaced TV material is why its happening...maybe most ppl are not seeing a problem on NTSC Progressive film encodes....Perhaps one of the interlaced frames is getting lost somehow, it looks like that, or perhaps that an interlaced frame is getting put back in the wrong order at the joins...??? But its very frustrating... :(

Blackout-

[robot1]

I wander if the fact that its a PAL DVD, and that its from Interlaced TV material is why its happening...maybe most ppl are not seeing a problem on NTSC Progressive film encodes....Perhaps one of the interlaced frames is getting lost somehow, it looks like that, or perhaps that an interlaced frame is getting put back in the wrong order at the joins...??? But its very frustrating... :(

Blackout-
I've rebuilt PAL interlaced DVD without problems.
Anyway, the frame order is handled by DVD-RB.
To check if it's a DVD-RB problem or a Procoder problem you should encode with a different encoder (there are free ones to test, as QuEnc or HC)
I could suggest also to upgrade to the last procoder version, as in the msg by daphy: it solves an issue with field order

[robot1]

The update for Canopus ProCoder express solves all the bug reported for the previous versions (2 frames missing in many encodes, average bitrate limited to a minimum of 2501 kbps) and add many parameters to tweak.
Now ProCoder express is a very good product for its quality/cost ratio, comparable to CCE Basic.

EclPRO v0.40 supports ProCoder express v1.12 (the older version isn't supported anymore).
It solves also the bug with NuMenu4U pointed out by D3s7

Download (as soon as a mod approves it)

[borgraf]

nice work once again robot1 :)

one question about a setting. I'd like to use GopLength 15 on all segments on a particular project, even the few above 3500 bps, but still have "AutoParameters" control the DC prec. Can AlwaysGOP15=1 be used to override the AutoParameters GOP control?

[robot1]

At present Autoparameters overrides the AlwaysGOP15 selection.
I will change the behaviour for next build.

[borgraf]

excellent :thanks:

[Rumbah]

Hi!

I want to use ECLPro for writing a Procoder Cuttermaran provider as I don't know the Procoder project format. Now I wanted to know what parameters from the ecl file ECLPro is using so that I don't have to write a complete eml file, but only the necessary information as I think not all things in there are relevant for Procoder.

Thanks

Rumbah

[robot1]

The ecl should be in the format used by CCE 2.50 or CCE 2.67.
These are the tags in the .ecl readed by EclPRO:


name=
aspect_ratio=
width=
height=
frame_rate_idx=
progressive=
top_first=
offset_line=
vid_file0=

vmode= or video_type= (one is used by cce 2.50 and the other from 2.6x)

according to the vmode:
cbr_brate=
vbr_brate_avg= and vbr_brate_min= and vbr_brate_max=
opv_brate_max= and opv_q_factor=


seq_endcode=
gop_hdr=
all_closed_gop=
intra_dc_prec=
seq_hdr=
encode_first=
encode_last=

[Rumbah]

Thanks for the fast answer.

I only have two additional question:

What values can video_type take and what is the meaning of them?
I couldn't find any information about it and if I create an ecl file with CCE SP 2.70.02 there is neither a vmode nor a video_type entry.
What q matrix does Procoder take? Does it take custom matrices from the ecl file?

[robot1]

vmode (CCE 2.50):
OPV 0
CBR 1
VBR 2

video_type (CCE 2.6x):
OPV 2
CBR 1
VBR 4


ProCoder uses his internal matrix, which can't be changed.
Q matrix parameters in the .ecl are ignored.

[FerretZone]

I just downloaded the Procoder Express 01.12.02 subrevision from Canopus to fix issues with the Athlon 64 processors on the nVidia 4 motherboards and EclPro 0.40 is not working. It is stopping and giving me an "Automation Error: Select Aspect Ratio". If I hit Ctrl-Alt-Del and then the "..." button for the Aspect Ratio, it clicks through and stops at the Job Summary window wanting me to hit the "Convert" button.

I am running an Athlon 64 3700 S754 on the MSI K8N Neo3-F with 2 Gb RAM and the ATI X700 PCIe w/256 Mb RAM (ATI branded). It was working with the 1.12 no subrevision, but the output was glitched like mad. I received the notice that the update was available from the Canopus tech support yesterday after sending them an email and downloaded the update today. That is when it broke DVD-Rebuilder Pro 1.00 RC2 and EclPro 0.40.

How do I fix this?

[FerretZone]

I ended up just putting PTFB <http://www.tlhouse.co.uk/PTFBPro/> on the system and have it pressing the "..." button after a 2 second wait (1 second didn't work) and the "Convert" button after 1 second. Everything is doing great with this cludge. I guess that is what happens when you put on the newest version of a program.

[djan]

Hi Robot1,

Could you explain each option available in the .ini file please ?

[Options]
Delay=0
Pause=10
Priority=0
Minimize=1
NoPathWarn=0
Procoder=C:\Program Files\ProCoder\Procoder2.exe
[Quality]
Quality=4
AlwaysGOP15=1
AutoParameters=1
[Debug]
StrictGOPBrateControl=0
Legacy=0 Thx.

[robot1]

- Options section
Procoder= Path to Procoder.exe
Priority = 0(default):Low - 1:Normal - 2:High
(with 0 you can surf or use the PC while encoding - encoding could be a bit slower, anyway it should be the best compromise)
Minimize = 0(default):Procoder window starts maximized while encoding - 1:Procoder window starts minimized

NoPathWarn = 0(default) - 1:"Target Path not set" warning will not be showed in non batch mode (DVD-RB uses it always in batch mode, so it's not important)

For ProCoder 1.5 or Procoder Express or Procoder 2 in legacy mode:
Delay = Delay in keypressing (default=0 - minimum)
Pause = TimeOut in waiting for windows to show (default=20 - suggested, you could lower to 10 with a fast pc)


- Quality section
Quality = 1:High Speed - 2:High Quality - 3(default):Highest Quality - 4:Mastering Quality (best quality)
AlwaysGop15 = 0:Set GopLength according to .ECL - 1(default):Always use GopLength=15 (better for low bitrates encoding)
Autoparameters = 0(default):Encoding parameters (DC Precision and gop length) are set according to the .ECL - 1:Encoding parameters are set by EclPRO, according to the bitrate to achieve maximum quality.
Alsaysframe = 0:Use field base encoding for interlaced sources (better) - 1(default):Always use frame mode (more compatible) - Please use frame mode in DVD-RB else you get errors in rebuilding. (currently it's hidden. You could use it if you remux with Scenarist: ex. in NuMenu)

- Debug section
StrictGOPBrateControl = 0:Default - 1:bitrate is controlled by each GOP unit. Use 1 if you have problem in your standalone player with default setting (should be not needed)
Legacy = 0:Used new ProCoder2 interaction (Default) - 2:ProCoder2 is used in legacy mode (commands sent by keys)
Legacy mode should not be needed anymore for ProCoder2...

[jdobbs]

@robot1

In case you haven't heard it lately -- you do good work!

[djan]

Thank you very much Robot1 !

Another question. Does ProCoder need ConvertToYUY2() ?

[robot1]

@jdobbs
:thanks:
Anyway, I can't express to you in my poor english all my thankfulness for making DVD-RB

@djan
ProCoder should work without ConvertToYUY2, but I don't know if it does the conversion internally, and in which colorspace it works.
Anyway, Avisynth does the conversion well, and I think it's safe to use.

[djan]

If I ask this question it's because I'm worry about the conversion produces a quality loss. And if it doesn't need the conversion, I prefer to not use it.

[wmansir]

robot1,
I know there is no trial version of Procoder Express, but I read that one comes with the trial version of Edius Pro 3, Canopus's video editing suite. According to the Demo download page the only limitation is:
> 30-day usage period
> TitleMotion Pro Not included

My connection is a bit flakey now, but I will download it when I get a chance and try it out.

[robot1]

Let me know the results.
While Express had many limitation before, now it's becoming a good alternative to CCE Basic for quality/cost ratio.
I would find a way to fully support Expess, without the "sendkeys"

[robot1]

If I ask this question it's because I'm worry about the conversion produces a quality loss. And if it doesn't need the conversion, I prefer to not use it.
Sorry for the late reply.
I can't answer to your question.
We should know how does Procoder work internally.
If it works in YUY2 workspace, the conversion in avisynth is harmless.
If it works in YV12, probably it could better if the conversion is omitted.

Anyway, I don't think it works in YV12... but it's just a guess.
Probably someone in the Other MPEG 1/2 Encoders (http://forum.doom9.org/forumdisplay.php?s=&daysprune=&f=62) could answer this question?