View Full Version : DGDecIM: Frame-accurate Avisynth source filter for QuickSync
Pages :
1
2
3
4
5
6
7
8
9
[
10]
11
jdobbs
19th April 2014, 23:40
Yes, sir. The default for fieldop is 0 anyway so you could omit it if that pleases you.Actually I'm more interested in using it for recovering film frames (on imported DVDs) -- so the fieldop option is great.
Thanks!
Guest
19th April 2014, 23:44
Actually I'm more interested in using it for recovering film frames (on imported DVDs) -- so the fieldop option is great.
Thanks! The concept of course is due to the legendary jackei of DVD2AVI fame. He also invented "force film". All hail jackei.
jdobbs
20th April 2014, 00:29
The concept of course is due to the legendary jackei of DVD2AVI fame. He also invented "force film". All hail jackei.I didn't know that. I second that "hail".
jdobbs
20th April 2014, 15:24
I'm getting decoding errors with pretty much any MPEG-2 (DVD resolution) source that I try to decode. The source video is remuxed into an M2TS using TSMUXER, and then is fed via AVISYNTH using DGIndexIM/DGDecodeIM. Here is the AVS file:LoadPlugin("C:\APPS\DGDecNV\DGDecodeIM.dll")
DGSourceIM("D:\WORKING5\WORKFILES\VID_00001.dgi", engine=2, fieldop=0)
ConvertToYV12().AssumeFPS(24000,1001)Below is a jpg showing the output (the source is showing a sky scene). The same source plays correctly when using DGDecodeNV. Other sources may differ somewhat, but all have issues within the first few seconds of playback.
Guest
20th April 2014, 15:48
I slipstreamed two fixes into b20 so re-download it if you need these fixes.
1. DGDecodeIM: Accept uppercase "DGI" extension without crashing.
2. DGIndexIM: Progress indication never goes backward.
Guest
20th April 2014, 15:50
What happens if you don't remux into M2TS, i.e., open the VOB directly with DGDecodeIM? Or is it that the M2TS plays correctly in NV but not IM?
Can you provide an M2TS fragment that I can use for testing?
jdobbs
20th April 2014, 16:06
What happens if you don't remux into M2TS, i.e., open the VOB directly with DGDecodeIM? Or is it that the M2TS plays correctly in NV but not IM?
Can you provide an M2TS fragment that I can use for testing?It plays correctly in NV but not IM.
I posted this in the wrong thread... I'll move it.
[Edit] I'll create a short fragment and get it to you.
Guest
20th April 2014, 16:17
OK, awaiting your sample...
jdobbs
20th April 2014, 16:28
OK, awaiting your sample...I sent a link via PM.
Guest
20th April 2014, 17:35
Thanks, investigating...
Guest
20th April 2014, 17:47
More shame for Intel...
It works just fine with the HW engine, but not for the SW engine. There is nothing I can do about it and asking Intel through their forum is a useless waste of time, as they only say they will ask their "experts" and then never come back with anything.
jdobbs
20th April 2014, 18:18
Ughhh... I guess I'll back-off on adding it to BD Rebuilder.
Thanks.
asteri
23rd April 2014, 09:09
I slipstreamed two fixes into b20 so re-download it if you need these fixes.
1. DGDecodeIM: Accept uppercase "DGI" extension without crashing.
2. DGIndexIM: Progress indication never goes backward.
I think the same bug I saw in DGDecNV, can you integrate "never go back" fix also in the DGDecNV ?
Guest
23rd April 2014, 14:32
At some point, yes.
MeteorRain
7th May 2014, 05:10
Recently I just tried the resize_h and resize_w parameters and it hangs after reading some frames.
EDIT: Being aware that I'm not on the latest version, I just update it to beta 20.
It appears that some videos get decoding smoothly (but incorrect image displayed), some crash on initializing.
Please use the latest version and provide a stream(s) to allow me to duplicate it.
MeteorRain
7th May 2014, 05:42
akiyuuki.dreamhosters.com/dgtest.tar
Sorry, I forgot that resizing is not implemented. I should add an error message. Thanks for pointing it out.
"It appears that some videos get decoding smoothly (but incorrect image displayed), some crash on initializing."
Do you have a stream for this and instructions for seeing the problem?
tormento
12th May 2014, 14:26
HW crop is not working yet, correct?
It does misunderstand dgi crop coordinates, correct?
Guest
12th May 2014, 15:21
Cropping and resizing are not implemented and I have to remove those parameters or add an error message. Further work on DGDecIM is low priority for me, however, at this time.
Nico8583
25th May 2014, 15:21
Hi,
Intel Media SDK 2014 R2 has been released on 12 May 2014 : http://software.intel.com/sites/default/files/MediaSDK2014R2Clients.zip
Perhaps it solves bugs (MVC Pacific Rim bug and some bugs found in this topic) ?
I'm trying to search changelog.
jdobbs
25th May 2014, 21:56
Hi,
Intel Media SDK 2014 R2 has been released on 12 May 2014 : http://software.intel.com/sites/default/files/MediaSDK2014R2Clients.zip
Perhaps it solves bugs (MVC Pacific Rim bug and some bugs found in this topic) ?
I'm trying to search changelog.I'm running "Pacific Rim" with FRIMDecode/FRIMEncode and the new DLL now. I'll let you know if it makes a difference.
Guest
25th May 2014, 22:07
Thanks, jdobbs! If it improves things, I'll update DGDecIM.
jdobbs
25th May 2014, 22:16
I ran it against the short test file that was created -- and it doesn't look like it helps. But just to be sure I'm now running against the original source disc. It will be a few hours (probably 5) before it completes.
Nico8583
26th May 2014, 10:52
Thanks jdobbs, I hope you'll get a better result...
Is it necessary to compile a new DLL ? Some functions have perhaps changed and need to recompile ? I don't know how Intel SDK works.
jdobbs
26th May 2014, 14:40
The new DLL is included in the package.
I ran "Pacific Rim" from the original disc. It completed, but still has the same issues at the same point as before.
Guest
26th May 2014, 15:40
FYI, there are parts of the SDK code that get compiled into the DGDecodeIM DLL; it's not all in the Intel DLLs. However, it's unlikely those parts would affect decoding in this case.
Nico8583
26th May 2014, 15:43
Thanks for information, so perhaps it is necessary to compile a new DLL with parts of the SDK code ? Or perhaps it could not solve the problem, even with this new version...
Guest
26th May 2014, 20:21
I don't think it matters. AFAIK, the only built-in piece is the distributor which determines whether to use HW or SW and then invokes it.
Carpo
28th May 2014, 21:36
Did a bit of testing
Intel Core i7 with Intel HD2000 graphics, using Intel 15.28.22.64.3517 drivers (these are the latest I can use, as newer version say my PC does not meat the minimum requirements)
Test 1 - DVD Source
AVS
LoadPlugin("D:\TestAVS\DGDecodeIM.dll")
dgsourceim("D:\TestAVS\DGIM.dgi",engine=1)
D:\TestAVS>avsmeter "D:\TestAVS\DGIM.avs"
AVSMeter v1.10 (Jul 17 2011 01:20:35) by Groucho2004
AviSynth 2.60, build:Sep 18 2013 [17:36:36]
Number of frames: 134798
Length (h:m:s.ms): 1:29:51.920
Frame width: 720
Frame height: 576
Framerate: 25.000 (25000 / 1000)
Progressive: Yes
Colorspace: YV12
Hit ESC to exit...
Frame 134798/134798, fps (min/max/avg): 725.23 | 1523.52 | 1063.21
TEST 2
AVS
LoadPlugin("D:\TestAVS\DGDecodeIM.dll")
dgsourceim("D:\TestAVS\DGIM.dgi",engine=1)
Load_Stdcall_Plugin("D:\MeGUI_2418_x86\tools\avisynth_plugin\yadif.dll")
Yadif(order=1)
AVSMeter v1.10 (Jul 17 2011 01:20:35) by Groucho2004
AviSynth 2.60, build:Sep 18 2013 [17:36:36]
Number of frames: 134798
Length (h:m:s.ms): 1:29:51.920
Frame width: 696
Frame height: 572
Framerate: 25.000 (25000 / 1000)
Progressive: Yes
Colorspace: YV12
Hit ESC to exit...
Frame 134798/134798, fps (min/max/avg): 579.17 | 850.80 | 727.12
As i understand cropping and resizing is not implemented at this time so I did not test with those, I am testing out the same script, but have added QTGMC, will post results later as it's taking a bit longer than yadif, which is to be expected.
QTGMC Result
AVSMeter v1.10 (Jul 17 2011 01:20:35) by Groucho2004
AviSynth 2.60, build:Sep 18 2013 [17:36:36]
Number of frames: 134798
Length (h:m:s.ms): 1:29:51.920
Frame width: 720
Frame height: 576
Framerate: 25.000 (25 / 1)
Progressive: Yes
Colorspace: YV12
Hit ESC to exit...
Frame 134798/134798, fps (min/max/avg): 6.04 | 7.28 | 6.83
Test 3 - Rat.264
AVS
LoadPlugin("DGDecodeIM.dll")
dgsourceim("rat.264.dgi")
AVSMeter v1.10 (Jul 17 2011 01:20:35) by Groucho2004
AviSynth 2.60, build:Sep 18 2013 [17:36:36]
Number of frames: 839
Length (h:m:s.ms): 0:00:34.993
Frame width: 1920
Frame height: 1080
Framerate: 23.976 (24000 / 1001)
Progressive: Yes
Colorspace: YV12
Hit ESC to exit...
Frame 839/839, fps (min/max/avg): 65.70 | 181.61 | 128.05
Test 4
AVS
LoadPlugin("DGDecodeIM.dll")
dgsourceim("rat.264.dgi", engine=1)
AVSMeter v1.10 (Jul 17 2011 01:20:35) by Groucho2004
AviSynth 2.60, build:Sep 18 2013 [17:36:36]
Number of frames: 839
Length (h:m:s.ms): 0:00:34.993
Frame width: 1920
Frame height: 1080
Framerate: 23.976 (24000 / 1001)
Progressive: Yes
Colorspace: YV12
Hit ESC to exit...
Frame 839/839, fps (min/max/avg): 68.34 | 182.66 | 128.62
Not the longest of clips I could have used I know, but it's all I have at this time, I will test with a few Blurays I have once the QTGMC test's have completed.
Nico8583
28th May 2014, 22:54
The new DLL is included in the package.
I ran "Pacific Rim" from the original disc. It completed, but still has the same issues at the same point as before.
Could you remind me position(s) bug(s) ?
Could you remind me position(s) bug(s) ?
I think this:
Look at 1:06:14. Soon after, you'll see the monster that appears in the fire. The bottom of the image is bad, then the whole image becomes bad. There are 6 bad frames in total.
First bad image in the sequence (Only the bottom part is bad):
http://imagizer.imageshack.us/v2/xq90/600/skjg.png (https://imageshack.com/i/goskjgp)
Last bad image:
http://imagizer.imageshack.us/v2/xq90/31/m367.png (https://imageshack.com/i/0vm367p)
Nico8583
29th May 2014, 18:07
Thanks, I have the same issue with my Pacific Rim (French Edition) :(
jdobbs
29th May 2014, 22:11
Yep, that's it.
Sharc
29th May 2014, 22:15
Still, Pacific Rim seems to be only remaining case causing this issue, right?
jdobbs
29th May 2014, 22:35
Still, Pacific Rim seems to be only remaining case causing this issue, right?it's the only one I've seen.
Nico8583
1st June 2014, 13:06
Does anyone tried 2D conversion with Pacific Rim ? Because the problem appears on AVC and MVC stream, so the problem is perhaps on AVC stream ?
Sharc
1st June 2014, 20:07
The independent stream (.264) plays without the glitches.
Nico8583
1st June 2014, 21:20
Have you played it with Intel Decoder ?
Sharc
1st June 2014, 22:15
with dgdecNV ......
Guest
1st June 2014, 22:17
That's inconclusive. Try it with DGDecIM.
Sharc
1st June 2014, 22:46
"This beta Version is expired" (?)
I am using b20 ...
Guest
1st June 2014, 23:01
Re-download from my web site.
Sharc
1st June 2014, 23:14
Independent (AVC) stream of Pacific Rim:
- Decoding with dgdecIM: Glitches
- Decoding with dgdecNV: No glitches
Guest
1st June 2014, 23:22
As expected. Now, try getting support from Intel. :devil:
Nico8583
2nd June 2014, 06:51
Thanks Sharc, so if we want to test (if Intel decides to provide support...) we can only play AVC stream. I'm going to try to open a case today and I hope Intel will take care of it...
MeteorRain
2nd June 2014, 07:31
DGIndexIM main window will get the input focus when starting, and even the window does not appear, press space will pause the progress, and if you ever lost the focus on the hidden window you'll have no way to resume the process, only Ctrl-C to terminate it.
Nico8583
2nd June 2014, 08:09
I have open this thread if anyone want to add/confirm informations : https://software.intel.com/fr-fr/forums/topic/516009
I have added a comment to your bug report, and I encourage everybody to do it too. If there are many complains, perhaps they will finally do something...
Nico8583
2nd June 2014, 10:52
Thanks r0lZ, I'm agree with you ;)
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.