Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
|
|
Thread Tools | Search this Thread | Display Modes |
28th April 2011, 01:07 | #61 | Link |
Registered User
Join Date: Sep 2010
Posts: 321
|
Portal 2! I already beat it in single-player mode, pretty fun puzzler.
__________________
MPC-HC/MPC-BE, Lav Filters, MadVR CPU: AMD Ryzen 5 1600, Video: AMD Radeon RX Vega 56 -> TCL S405 55", Audio: Audio-Technica M50S |
30th April 2011, 09:19 | #64 | Link |
Registered User
Join Date: Apr 2009
Posts: 478
|
Thanks Subjunk! I tried Interframe for one whole DVD yesterday, which annoyingly was not interlaced, thus I couldn't use QTGMC to bob it. Interframe worked very well. I mean, quality and speed, what more could I ask for?
I'm sure the purist will say that framedoubling is flawed, but hey I want 59.94 fps and Interframe is great for that, so why not? |
30th April 2011, 22:22 | #66 | Link | |
Registered User
Join Date: Mar 2010
Posts: 5
|
i have this error u.u
me script Quote:
|
|
30th April 2011, 22:25 | #67 | Link | |
Registered User
Join Date: Jun 2010
Posts: 443
|
Quote:
It's because your FlowPath should be the path, not including the filename. See the help page in the download, or online here for details on each parameter |
|
1st May 2011, 22:25 | #69 | Link |
Registered User
Join Date: Apr 2002
Location: Germany
Posts: 5,389
|
One sample is linked in post#1.
InterFrame is there to be used. So, just use it.
__________________
- We´re at the beginning of the end of mankind´s childhood - My little flickr gallery. (Yes indeed, I do have hobbies other than digital video!) |
3rd May 2011, 21:52 | #71 | Link |
Registered User
Join Date: Jun 2010
Posts: 443
|
It really shows how juddery regular videos are; it's not the script that makes it juddery, it's just that the rest of the frame is improved while the cigarette remains like the original video
However, you notice things like that much more when the interpolation is still a novelty; when you watch it all the time you get used to it Last edited by SubJunk; 7th May 2011 at 02:18. |
8th May 2011, 17:36 | #73 | Link |
Registered User
Join Date: Apr 2002
Location: Germany
Posts: 4,926
|
Its not only the cigarette but i have the feeling something is wrong with the background objects that passing by, might be just imagination though but i have the feeling watching the sample something is wrong.
Did someone compared this allready vs other FRCs like Yuvsoft, MSU or Cyberlinks Tzan (CUDA implementation) ?
__________________
all my compares are riddles so please try to decipher them yourselves :) It is about Time Join the Revolution NOW before it is to Late ! http://forum.doom9.org/showthread.php?t=168004 Last edited by CruNcher; 8th May 2011 at 17:42. |
8th May 2011, 17:52 | #74 | Link |
Registered User
Join Date: Apr 2002
Location: Germany
Posts: 5,389
|
Doing a non-integer framerate interpolation (like in 24->60) is always visually more problematic than doing an integer conversion (like in plain framerate doubling). When doubling, every second frame is original, and the potentially artificial/or/inconsistent interpolations show up only a short time. When you do 24->60, then you have one original frame, then come 4 (four!) interpolated frames, one original, 4 interpolations, etc. That's *much* more time for (potential) artifacts/inconsistencies to show up and be noticed.
__________________
- We´re at the beginning of the end of mankind´s childhood - My little flickr gallery. (Yes indeed, I do have hobbies other than digital video!) |
8th May 2011, 18:04 | #75 | Link | |
Registered User
Join Date: Apr 2002
Location: Germany
Posts: 4,926
|
Very well explained Didée thx if i understand it right this is once again combining needi + mvtools to achieve this so we have now FRC,Deinterlacing,Denoising,Deflicker,AddGrain,Degrain all based on mvtools as the motion estimation engine and needi playing a big part too that's great and for everyone we can partly utilize the GPU as well . And by combining ffdshow with this we can even utilize it in Realtime frameworks and with DSP support (supplying the frames for decoding leaving more CPU for the PostPro)
The SVP guys actually seem to make the first steps into integrating this in a easy to use Realtime package absolutely great Im really wondering how avisynth and it's framework by these days compares vs high class stuff like thefoundry or greenparrotpictures Research and YuvSoft also isvs like Cyberlink Surely avisynth didn't reached 10 Bit yet but i wonder how the base algorithms compare and where it stands Like there FRC http://www.greenparrotpictures.com/slomo.php or http://www.greenparrotpictures.com/flicker.php though also mostly impossible to compare i guess with those without having access to them Quote:
Here is the actual product where their stuff gets into action and what would be comparable to Interframe http://www.thefoundry.co.uk/products/kronos/ http://features.cgsociety.org/story_...?story_id=5776 (the move from C++ to GPU) here is a moving demo of the scene in the cgsociety article http://www.youtube.com/watch?v=hsFKiLxC4js Especially the scene is interesting most of their Demo scenes they show Watter (chaotic motion) the canoe paddle shows though some heavy estimation problem, their solution though seems not to apply for interframe in the cigarette case or passing background http://www.youtube.com/watch?v=L0Dl8wyBolQ Also it seems that their GPU version doesn't show this issue anymore and conquers it now in real time (looking @ the cgsociety demo of the same scene) without needing to apply any alpha matte anymore to fix this Hey Hey Hey http://thefoundry.s3.amazonaws.com/d...s_Media-v4.zip the assets of that tutorial the Watter scene http://www.youtube.com/watch?v=tWWGqQImje4 <- Motionblur assets http://thefoundry.s3.amazonaws.com/d...ming_Media.zip <- seems damaged though http://thefoundry.s3.amazonaws.com/d...Blur_Media.zip <- Motionblur assets in .dpx http://www.youtube.com/watch?v=YPLj5pQmbV8 <- Grain and Denoise assets http://thefoundry.s3.amazonaws.com/d...rain_Media.zip The Tutorial Videos are also available as higher quality Quicktime movs not perfect for comparing final results but better then with those youtube videos http://www.thefoundry.co.uk/articles...-nuke-denoise/ http://www.thefoundry.co.uk/articles...r-nuke-kronos/ http://www.thefoundry.co.uk/articles...ke-steadiness/ Also even more assets to play with http://www.thefoundry.co.uk/products...core/training/ The finalcut assets are either in mjpeg 4:2:2 or raw or rle PS: Can somebody decipher what for a Xeon CPU they compare with http://media.cgnetworks.com/flash/foundry/hawaii.swf Intel Xeon ES5C4 or E55C4 hmm its like Google doesn't know it ?
__________________
all my compares are riddles so please try to decipher them yourselves :) It is about Time Join the Revolution NOW before it is to Late ! http://forum.doom9.org/showthread.php?t=168004 Last edited by CruNcher; 8th May 2011 at 23:01. |
|
8th May 2011, 23:35 | #76 | Link | |||
Registered User
Join Date: Jun 2010
Posts: 443
|
Quote:
Quote:
There are definitely errors in the sample, and in fact that sample hasn't been updated in a while so I should remake it as there have been quality improvements since then. Of course it still won't be perfect Quote:
I emailed the YUVsoft and MSU guys but they said they don't have products for individual people, only for companies. I've also tried a few more products and found them to either be worse or had other problems (for example one of them only works with .avi files so I didn't bother trying it) I'll read through your big post now and reply later. It's great to have all this feedback in the thread, maybe we can improve the script with all these eyes on it |
|||
9th May 2011, 01:01 | #77 | Link | |
Registered User
Join Date: Jun 2010
Posts: 443
|
Quote:
I should also note that InterFrame is not meant for slow motion and there is a change I would make in the code if it were meant for that. It may seem like framedoubling and slow motion are the same thing, and on paper they are, but our eyes see things differently depending on whether they are moving fast or slow so that is where the difference is. For example slight blurring and ghosting can look good at fast speed while in slow motion it looks bad, so the technique is similar but not the same. Samples: GreenParrot - InterFrame. P.S. I still have another reply to do, but I'm working so that's why I'm doing separate replies in my breaks Edit: The Kronos trial installation doesn't work for me so I can't test it. Last edited by SubJunk; 9th May 2011 at 03:19. |
|
9th May 2011, 03:58 | #78 | Link |
Registered User
Join Date: Apr 2002
Location: Germany
Posts: 4,926
|
looks very good except that speed ramp or artifact @ 0:06 but taking into account you worked with the compressed youtube source it's identical in terms of motion feel, surely this way it can't be nicely compared in other parts like interpolation quality and speed
http://thefoundry.s3.amazonaws.com/p...release-64.zip <- for everyone who wants to give it a compare shoot,dont forget it's an After Effects plugin http://www.thefoundry.co.uk/products...-requirements/ PS: It looks like without the SVP team these speeds wouldn't have been possible @ all amazing work from Russia again a whole OpenCL implementation for the Optical Flow
__________________
all my compares are riddles so please try to decipher them yourselves :) It is about Time Join the Revolution NOW before it is to Late ! http://forum.doom9.org/showthread.php?t=168004 Last edited by CruNcher; 9th May 2011 at 04:47. |
9th May 2011, 08:40 | #79 | Link | |
Registered User
Join Date: Jun 2010
Posts: 443
|
Quote:
The upcoming version increases speed even more |
|
13th May 2011, 09:13 | #80 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Hey SubJunk,
what kind of motion estimation technique are you currently using? Have you thought about using Optical Flow + Warp, if you're not using that yet? Optical Flow algorithms seem to have improved quite nicely. They're still rather slow (at least the good ones), but I'd guess that they should allow you to improve quality quite a bit. Here are some links to open source Optical Flow algorithms: http://people.csail.mit.edu/celiu/OpticalFlow/ http://www.onera.fr/dtim-en/gpu-for-image/folkigpu.php http://sourceforge.net/projects/open...ptical%20Flow/ http://nghiaho.com/?page_id=189 http://plyer.fr/aurelien/?p=124 http://varflow.sourceforge.net/ http://perception.inrialpes.fr/~chari/myweb/Software/ Here's the "official" Optical Flow benchmark site. It seems that there are rather large differences between different Optical Flow algorithms: http://vision.middlebury.edu/flow/ev...results-i1.php Interesting approaches would also be to compare motion vectors between consecutive frames and only use those which are comparable. That should nicely sort out erratic motion vectors. Finally, how are you filling the occluded space? If you check motion vectors to the next "future" frame, too, you could eventually fill the occluded space from the future frame. That might help removing many artifacts. Just shooting from my hip here, haven't really done any motion estimation programming myself yet. Just some brainstorming from what I've read... Last edited by madshi; 13th May 2011 at 09:29. |
Tags |
48fps, 60fps, framedoubling, interframe, smooth motion |
Thread Tools | Search this Thread |
Display Modes | |
|
|