![]() |
@kwag and jellygoose: I'm using the same trigger value for 480x480 encodes and it works just fine.
|
Quote:
btw: Where did STMedian Filter go in the latest script :?: |
Quote:
-kwag |
Quote:
Too many cycles are wearing my brain out :P -kwag |
Quote:
i see without STMedianin in the last and before the last too but don't post. i was afraid to bore with some that i don't understand. :oops: thank you all. :wink: doing the sample number 39738390,8 :lol: |
i did samples with and without the (forgoten)STF and can't see differences.
with the filter i got little less size, but in pc the quality is the same. anyone did the same? see differences :?: in the last news i can't see some strange artefacts in the moviments on the background using 480x480. is really better :!: |
[quote]
Quote:
|
[quote="ovg64"]
Quote:
|
Now that the dust has settled after evryone's hard work I thought I might share my experience for adding filters for not so clean material/captures.
UnDot and DNR2 do a good job of removing noise and are a balance of speed and quality. Actually, Kwag, you might want to consider adding UnDot to the current script for DVD's since it has virtually no effect on encoding time. |
Right now, all I can say is that the latest script still works best for me, but I'm currently testing aSharp Filter, which is an Adaptive sharpening Filter, and the results are promising...
Testing is such a long-taking procedure for me... :wink: |
Which Sharpen Filter, JellyGoose :idea: :?:
EDIT: Now, I read that you're testing "aSharp" :oops: |
Quote:
asharp is the best sharper that i know. i'm doing samples for 2 days to see and compare using in the Kwag's scripts(and i test all scripts after any change) changing unfilter to asharp and compare any encode. in all asharp is champion! do a test and see :!: asharp(2,3)...(very gently) or asharp(2,4)...(more sharp) :wink: |
Sounds very Promising, Jorel :!: :D
Am I supposed to put "aSharp" Before(Like UnFilter) or After resizing :?: :wink: @Kwag It's time for updating the script... ...again :D |
Quote:
all "sharp" filter tested before resize give more sharpness and less size...ever :!: :wink: |
Yes I know, Jorel :wink:
I just required your opinion as final confirmation :) I just made my first sample and I'm very curious... :lol: |
Hi all if this is thru than its worth using asharp. :D I have tried it and so far so good, but keep in mind the filter is very strong and at a high setting will add file size.
What it does: basically, this filter is a very common unsharp mask, simply because unsharp mask is the most pleasant sharpenning technique for my eyes ^^ but i've added 3 simple ideas i had about sharpening : - adaptive thresholding to avoid noise enhancement - block adaptive sharpenning to avoid DCT block edges enhancement - block based adaptive thresholding to avoid DCT block edges enhancement |
Hmm... This ASharp filter sure sounds interesting! What settings do you guys recommend for a medium to high action DVD movie? I'm currently encoding "A Knight's Tale" and it would be nice to test the filter on that movie! Do you use Bicubic or Linear resize together with ASharp?
|
Quote:
i got a little less CQ after prediction but big more quality and of course,after prediction the final size will be the same using any filter. :wink: audioslave, the recomendation from the developer is asharp(2,4) but it will encrease the size and the Kwag's scripts, don't need huge sharpness. better is (2,2) or(2,3). see how this filter encrease the contrast :!: :wink: ps: where is our master? (Kwag of course) :?: |
@jorel
What about the other two parameters? No need for them? And again: :wink: Do you use Bicubic or Linear resize together with the filter? |
@dano,
Thanks for the tip on "undot". It's now in the script. @jorel, Thaks for your heavy testing with "asharp". I've removed "unsharp" and used "asharp". @all, Thanks for your patience :D I've re-written the script, eliminating ALL dynamic blur and mergeluma(blur) calls. I've used negative values in asharp to produce exactly the same dynamic results, but without the DAMN bugs that the "blur" and family of functions. I had to do this, because I still found sporadic flashes on several encodes I did today. So I tried asharp, but used it as a "blur" engine instead of as a sharpener. Only at the beginning of the script it's used to oversharpen, just like unfilter was used. Also, the speed of the script is much faster by removing all the 'blur" functions. So give it a whirl, and post your comments. I can now say that I didn't find ANY blink or flash anywhere on scenes that DID have flashes with the previous script. The functionality of the new script is identical to the previous one. -kwag |
I'm currently trying out the asharp's adaptive sharpening. If it works as advertized ( 8) ), then I'll turn it on. From the readme file of asharp, here's what it does:
- adaptive thresholding to avoid noise enhancement - block adaptive sharpenning to avoid DCT block edges enhancement - block based adaptive thresholding to avoid DCT block edges enhancement So this combination, together with the adaptive filtering (bluring) on action and scene changes, should be a killer :D -kwag |
Note: The first line of "asharp" in the script should read: asharp(2,0) and not asharp(1,0).
I fixed that. :imstupid: |
@Kwag
you wrote: "@all, Thanks for your patience" are you kidding? we all are thankfull for your great work. :D without your friendship and developing, this forum couldn't never be so great. thanks for the script update too. :wink: @audioslave i ever use bicubic precise in all my tests and in the last encodes. :) @all great work! i'm thankfull :D the Kwag team is the best. :!: |
MY Test
I just did a couple of tests using asharp.
Test Basics: (Current Script) kvcdX3 - MPEG1 CQ=80 Sample Length=1656 Frames (69 seconds) MinBitrate=300, Max Bitrate=2500, VBV=40, GOP 1-5823-2-1-24 Filesize with asharp(2,0) = 8,653,029 Filesize with asharp(2,2) = 7,236,833 (16.37% smaller) kvcdX3 - MPEG2 CQ=80 Sample Length=1656 Frames (69 seconds) MinBitrate=600, Max Bitrate=6000, VBV=224, GOP 1-5823-2-1-18 Filesize with asharp(2,0) = 19,076,155 Filesize with asharp(2,2) = 15,308,597 (19.75% smaller) I don't see much difference in quality (but I don't have the best eyes. :) ) I think that adaptive shapening is the way to go with asharp. Racer99 |
Here's a small ( ~1MB ) sample with the new script :D
www.kvcd.net/red-planet-short-asharp-704x480.mpg -kwag |
Re: MY Test
Quote:
-kwag |
:imstupid: :imstupid::imstupid::imstupid:
Latest correction: I was passing bluring variables as INT instead of FLOAT :imstupid: The scaling on blurring was incorrect, as it was in harder steps ( -1, -2, -3, etc. ). Fixed. It's now floating point values ( -1.12, -2.142, etc. ) -kwag |
hey Kwag,
i don't test the "very new" yet, i see it only now. the last script before your corrections give some staircases like in lanczos, did you see some like this? i see it in detail in 352x240. with more "resize" this thin lines with staircases are hard to see. can i encode a movie using 720x480 in one cdr80 or cdr90 with audio in 160k and got good result :?: i'm asking cos never did with more than 480x480. i did only samples in pc and the results are incredibles. i never see this resolution in the tv. ps: i got a new player phillips and he play all kvcds. 8) i did a big search to find what was the best. 2 weeks testing with tons of kvcds in the shops, 8O i post it in the portuguese forum. the important now isn't the dvd player. the best is the kvcd player :!: :wink: @racer, i'm with you! asharp(2,2) ... (2,4) give more sharp and less size! i think that (2,2) is a "gently" sharpness. :wink: |
Quote:
The "Red Planet" samples I posted were all 704x480 with a CQ for one 80 minute CD :!: Audio at 112Kbps. -kwag |
Quote:
what CQ did you find in the sample :?: |
Quote:
-kwag |
God I was testing the whole night, and now everything has changed again... :cry: 8)
well as for noise reduction and sharpness, I found that this line would best replace Unfilter(50,50) : Code:
Unfilter(10,10) |
Quote:
Did you see the latest one: asharp(2, 4, 0.25, hqbf=true) Have you tested that one :?: It seems that asharp is way faster than unfilter :!: -kwag |
currently testing... :wink:
|
Quote:
-kwag |
@kwag
So what are you waiting for?! Show us a clip, man! :D EDIT: Would be nice with info on CQ etc. |
Quote:
-kwag |
:lol: :lol: :lol:
|
Ok, this one should top just about every other sample ever made at KVCD.Net :mrgreen:
www.kvcd.net/red-planet-extreme-704x480.mpg Specs: 31 second sample, 3,838KB. Res. 704x480, CQ=63.42, MIN bitrate 300Kbps, MAX bitrate 2,500Kbps KVCD "Notch" Q. Matrix. GOP 1-12-2-1-24, Current script as posted Ver. June 5, 2003 @05:45GMT :mrgreen: -kwag |
@kwag
I didn't have any luck with squeezing "A Knight's Tale" onto one CDR :cry: . When I made the prediction I got a CQ value 25-30(!) for one CDR, and I guess that won't be very exciting to look at... If a movie contains a lot of fast action scenes will it be easier to fit on a single disc? |
Site design, images and content © 2002-2026 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2026 Jelsoft Enterprises Ltd.