MSU Lossless Codec 0.6.0 released!

[DmitriyV2]

MSU Lossless Codec 0.6.0 released!

Main advantages:
* New compression algorithm implemented. Compression speed increased 1.5-2 times, decompression speed increased 2-2.5 times!
* New interface: more simple and clear.
* Can decompress files created by 0.5.x version, can compress with the old algorithm also.

http://www.compression.ru/video/ls-codec/index_en.html

Enjoy! ;)

[Inc]

IMHO your link seems to primary point on your Deblocking filter (00:47 am) ;)

http://www.compression.ru/video/ls-codec/index_en.html

[video_magic]

Thankyou M.S.U. Team :) I look forward to checking this out for my next lot of captures, I will be comparing it to Arithyuv. I have a BT878a based card.

[Sharktooth]

website went down while i was browsing it :(

[DmitriyV2]

IMHO your link seems to primary point on your Deblocking filter (00:47 am) ;)
Absolutely! ;) Thanks!

[DmitriyV2]

website went down while i was browsing it :(
Check now please!

[movax]

Ended up a gig smaller than a loco/vble version. More HDD space for TV rips. :)

[Sharktooth]

Check now please!
Thanks! It works perfectly now. Just trying the codec :)

[johnsonlam]

MSU Lossless Codec 0.6.0 released!

Main advantages:
* New compression algorithm implemented. Compression speed increased 1.5-2 times, decompression speed increased 2-2.5 times!
* New interface: more simple and clear.
* Can decompress files created by 0.5.x version, can compress with the old algorithm also.

http://www.compression.ru/video/ls-codec/index_en.html

Enjoy! ;)

Me again!
Tonight I'll test it, thanks DmitriyV2! (still the same guy in Neuron.net?)

[DmitriyV2]

Me again! Tonight I'll test it, thanks DmitriyV2! (still the same guy in Neuron.net?)
The same! :) Please let me know, if any problems with tests!

[johnsonlam]

The same! :) Please let me know, if any problems with tests!

Thanks! Good speed, easy interface and accurate result (I use 100% loseless).

But a minor problem:

Play an incomplete AVI encoded with VirtualDub with Media Player Classics, (seems the only application can handle MSU codec). MPC will crash, pop up the following MSUlvc exception windows:

CodecInst::DecompressEnd
CodecInst::DecompressBegin
RangeCoderV::init
RangeCoderV::renew called no init
DecodePlane x2
Upredict2 x2
DecompressIFrameRGB

Now click any button of MPC will crash MPC with a "ntdll.dll" error.

Maybe the only thing MSU Lossless Codec need to improve is error handling and directshow filter.

Really thanks a lot, you guys are always making great filters and fair reports!

[ReinerSchweinlin]

Thanx for this nice codec!! Gives very good compression results. I'm doing some testing to see if it´s suitable for archiving :)

[Igor Kharchenko]

Thanks for codec and different useful tools!
Do you have any plans for making DirectShow version. When I try to capture TV with FLY2000TV using vfw MSU codec (768x576 resolution) I have too many drop frames. With PicVideo MJPEG 3 (DS) I have no problems.

[DmitryPopov]

Exception boxes may appear when video image size is not a multiple of 16 (i.e. not 16*n).

Frames dropping is not related to DirectShow, the codec is just still very slow. You can achieve realtime with pictures like 352*288 (especially with Delta-Frames disabled), but not with 768x576.

By the way, disabling Delta-Frames gives speed boost and often makes compression better.

[Elic]

My problem is: when I start encoding (VirtualDubMod, Fast recompress, source is Avisynth with YV12) I get a message box "More data is available" and then codec encodes with 0.2 - 1 fps almost independent of picture size. What can I do for more precise diagnostics?

[DmitryPopov]

Please download the latest version from site (compression.ru/video/ls-codec). No "more data is available" anymore. ;)

[bananix]

i can't play video with MPC or open with virtualdub that i compressed with this thing :(

EDIT: actually i can with one i compressed with it...perhaps the resolution needs to be multiple of 16 or something...but i convert quicktime clip with RAD Video Tools and it has some option to change resolution...but don't know can it be largened to add black borders...

[bananix]

the bink converter accepted negative values to the crop fields and so letterboxed it to 640x480 and then it played...

[Chainmax]

I trying this to create an intermediate file that will be fed to CCE. I use quality=absolute lossless and mode=max compression, slow decompr., but when I click on the Advanced tab to select motion estimation=maximum compression, mode gets switched to manual tuning (advanced button), which I assume makes use of the denoising settings. How can I disable denoising?

[foxyshadis]

Setting 0 in both denoising boxes. Switching presets won't change what the manual mode uses, so you'll have to set all the ME manually.

[WorBry]

I really quite impressed with the denoising component of the codec. For my DV sources I was using the following 2 stage process to convert 25i to 25p for archiving as a YV12 ''lossless'' master (used primarily for conversion to MPEG-4)

Stage 1

a) Decompressor: Cedocida DV codec, with forced YV12 and 'MPEG2 interlaced' YV12 chroma sampling

b) AVS Script (Bob-Deinterlace & Denoise)

avisource("Type II DV.avi", pixel_type = "YV12")
Tweak(coring=true)
AssumeBFF()
mvbob()
LanczosResize(720,544)
Deen("a3d",3,3,0,3)
RemoveGrain(2)
OR
LRemoveDust_YV12(17,2)

c) Output: Fast Recompress, FFDShow HuffYuv YV12, Median Predictor, Huffman Tables

Stage 2

a) AVS Script: (50p to 25p filter with motion blur, based on Scharfis-Brain's mvfpsscd function using MotionProtectedFPS instead of mvfps)

avisource("HuffYuv YV12 50fps.avi")
mpfpsscd(25,oversample=16,blurradius=4)

function mpfpsscd(clip i, float fps, int "oversample", int "blurradius")
{
oversample=default(oversample,1)
blurradius=default(blurradius,1)
mv=i.MotionProtectedFPS(fps*oversample,warpfast)
global mv= (oversample>1) ? mv.temporalsoften(blurradius,255,255,mode=2).selectevery(oversample,0) : mv
global ch=i.changefps(fps) #. subtitle("ch")

global msk0=ch.duplicateframe(0).reduceby2().reduceby2().motionmask(thsd=255,y=3,u=1,v=1)
global msk1=msk0.trim(1,0)
global msk2=msk0.trim(2,0)
global msk3=msk0.trim(3,0)

f0=scriptclip(ch,"""sc""") #.subtitle(string(a0)+" "+string(a1)+" "+string(a2))""")
f1=f0.frameevaluate("""sc= (sc2 || sc1) ? ch : mv""")
f2=f1.frameevaluate("""sc2=( (a2 > (a1*5+25)) && (a2 > (a3*5+25)) ) ? true : false""")
f3=f2.frameevaluate("""sc1=( (a1 > (a0*5+25)) && (a1 > (a2*5+25)) ) ? true : false""")
f4=f3.frameevaluate("a3=averageluma(msk3)")
f5=f4.frameevaluate("a2=averageluma(msk2)")
f6=f5.frameevaluate("a1=averageluma(msk1)")
f6.frameevaluate("a0=averageluma(msk0)")
}


function MotionProtectedFPS(
\ clip input, float FPS,
\ int "mode", int "protection",
\ int "iterate", int "reset", int "initialise")
{
mode = default(mode, warpfast)
protection = default(protection, 80)
iterate = default(iterate, 2)
reset = default(reset, 50)
initialise = default(initialise, 6)

originalFPS = input.framerate()
input

mp = FindMotion(iterate = iterate, reset = reset,
\ initialise = initialise)
p = MotionFPS(FPS,mode,Motion = mp)
maskp = mp.SumStretchFPSMask(FPS, protection)

mn = FindMotion(iterate = iterate, reset = reset,
\ initialise = initialise, from = next)
n = MotionFPS(FPS,mode,Motion = mn, source = next)
maskn = mn.SumStretchFPSMask(FPS, protection, source = next)

maskp
assumeFPS(originalFPS)
CombineFPS(FPS, maskp, maskn.Invert())
bilinearresize(width*8, height*8)
crop(0, 0, input.width(), input.height())
fity2uv()

maskedmerge(p, n, last, u=3, v=3)
}

b) Output: Fast Recompress, FFDShow HuffYuv YV12, Median Predictor, Huffman Tables

For Stage 1 output I've now switched to MSU Lossless 0.6 with the following settings:

Compression Quality: Allow denoising (visually lossless), Manual Mode
Advanced:
Comp Method: 0.60
Denoise: Default Temp 30 Spatial 4.00
No Delta Frames
Motion Estimation: Full Search, PP-Frame

Using the denoise features of the MSU codec in this way, I've been able to achieve a compression boost (after final conversion to DivX 6) approaching that achieved with the primary denoise filters that I had used the the Stage 1 script, but (IMHO) with better preservation of fine detail and less ''dull'' smoothing. As such, I've now mostly dropped the AVS denoise filters, with the exception of RemoveGrain(1,2 or 5), which I may leave in for noisier 'low light' shots.

However, I'm wondering - does the codec denoise in the input color space (in this case YV12) or is there internal conversion to RGB for the filtration? I ask this because I see that the denoise filter has now become available as a separate VDub plug-in, which I assume operates in RGB.

For now, I'm still using FFDShow HuffYuv YV12 for the final (Stage 2) 25fps master, primarily for the decompression (playback) speed, but I'll play around some more with MSU.

Thanks for a great codec and denoiser.

[DmitryPopov]

In the codec all denoising is done in YV12. If the input is in YV12, denoising happens in original colorspace. If the input is RGB24/32 or YUY2, it is first converted to YV12 and then denoised in YV12. All denoised material is stored in YV12.

[WorBry]

Great, thanks.

[WorBry]

Setting 0 in both denoising boxes. Switching presets won't change what the manual mode uses, so you'll have to set all the ME manually.

It seems that the Spatial Denoising Intensity setting doesnt allow you to enter 0 directly - throws up the error message ''Intensity must be a positive real number''.

A way around this is to enter 0.0001 which is then accepted (once only) as 0.000, but it would be helpful to be able to enter 0 directly or else have a check-box for 'no denoising' that sets these values to 0.

Thanks

[ReinerSchweinlin]

Just to re-check....

On my Athlon XP @2000Mhz, I get compression fps lower 1fps, doing a "fast recompress" of a 640 x xyz xvid encode of a TV-capture. Is that normal? Settings are to max compression

[DmitriyV2]

Just to re-check....
On my Athlon XP @2000Mhz, I get compression fps lower 1fps, doing a "fast recompress" of a 640 x xyz xvid encode of a TV-capture. Is that normal? Settings are to max compression
Max compression is the slowest preset.

recently we release MSU Screen Capture Lossless Codec with another several times faster method. Now we workin on combine method to provide faster decompression and faster compression.

[ReinerSchweinlin]

Thanx for the reply. Is <1fps reasonable?

[DmitryPopov]

I guess, <1fps is when FullSearch option is turned on. This mode is for maniacs aiming to reach maximum possible compression. ;) For practical use FullSearch should be turned off. The speed will increase significantly, though still not realtime. Also, play with other options. Sometimes v0.6.0 compresses better with delta-frames turned off, when the speed is very good.

[DmitryPopov]

it would be helpful to be able to enter 0 directly or else have a check-box for 'no denoising' that sets these values to 0.

On the main configuration page there is an option called "Quality". If it isn't set to "Allow denoising" or "Allow stronger denoising", then no denoising happens no matter what denoising parameters are set in the "Advanced" window.

[WorBry]

Thanks Dmitry. Yes, I realized that after I last posted and did some quality metric comparisons.

[adoniscik]

I get DecodePlane errors followed by a jumbled mess (noise) when I try to import MSU LC encoded clips into After Effects 7.

[foxyshadis]

On the configuration tick "Always suggest RGB for output", because all adobe software is 100% RGB-only.

[johnsonlam]

Max compression is the slowest preset.

recently we release MSU Screen Capture Lossless Codec with another several times faster method. Now we workin on combine method to provide faster decompression and faster compression.

MSU Lossless Codec is good, but still have a few "rare" crash happen during capture or process with Virtualdub, looking forward to next version that can completely replace HuffYUV.

Thanks.

[adoniscik]

On the configuration tick "Always suggest RGB for output", because all adobe software is 100% RGB-only.
Close, but no cigar. Now it does not give any error message but it still displays a corrupted screen. I will avoid this codec for now :(

[DmitryPopov]

Try disabling key frames in codec's settings, it can help.

[adoniscik]

It appears you are right; it works now.

[DmitriyV2]

It appears you are right; it works now.
Good. Thanks for confirmation!

[nicksteel]

When using MSU Lossless Codec with CamStudio, what are best settings for CamStudio?

Trying to capture an aquarium screensaver (DreamAquarium), full screen. My machine, Dual-Core with Nvidia 7600GT will capture smoothly with Fraps at high resolutions, but CamStudio seems capture better images. Can't figure out best settings for smoothness with CamStudio. (I will process to DVD with avisynth, so need best original capture.)

[DmitryPopov]

This codec is very slow and cannot be used to capture full screen with good framerate. It was created to be used for archiving when compression ratio is the most important thing.
For your task try a faster codec like huffyuv.

[nicksteel]

Would Huffy be faster than CamStudio's codec? Also, my biggest problem is setting the CamStudio parameters for:

Set the Capture Frame
Set the Set Key Frames
Set Playback Rate

My final speed will be 29.97. Need highest quality settings, assuming use of CamStudio's codec.

[foxyshadis]

Since you have dual-core, try Lagarith. Single threaded it's quite slower than huffyuv, but dual-core it's nearly as fast. Huffyuv is the gold standard for speed, but takes the most space.

Camstudio codec is more like a less efficient MSU Screen Lossless, it's only really good at handling low-complexity screens that hardly ever change much.

[video_magic]

I just wondered if this was going to be developed anymore, I noticed the version is still at 0.6, but assume other lossless codecs like UTVideo have made it redundant?

[DmitryPopov]

Sorry, this project is discontinued and not supported anymore.
Thanks for mentioning Ut Video, didn't know about it before.