@tosvus I have a wide internet connection, so I can download any file size. You can use a file store services: Google Drive, Dropbox, etc. to share it. They are fast.

@spacewig a help will be published online, because also big examples will be published there to demonstrate "how-to". I have no plan to include it to a distribution. Currently no help available yet. I am focusing now to a program stability. Also I am testing it currently on a real project before publish.

@MikhailK. Wrong way at all...

"In x64 directory I renamed x264 to x264-10bit.exe" - the program will not work with a renamed 8-bit x264. Only a real 10-bit version should be used. Also x264 10 bit is already included to the coming beta (and latest alpha, as I remember). No any download and rename required to make 10 bit results.

Your steps are correct in generally, but you forgot to execute "Make.bat" files in every step (or if you use a very old alpha - "Apply"). They are necessary. This trick is a speed optimization to control every step with a comfort fps, because a lot of CPU needed to whole processing chain.

This may look complicated, but if you have a lot of long files, you get the fastest and comfort way to process them. The program is optimized for very long batch processing with many processing hours in every step.

Some important knowledge things:

An Avsi file is a project file. It contains a list of all step parameters and filters used in every step. So when click to "Add to/Make Avsi" you include these steps to that file. Also you had no any tuning steps, as I see, so your result will contain an unknown "default" set of filters. So you will get an unknown processing and an unknown result. Because no any video file extracted and created in every step (no Make.bat had executed), you had only a short black windows. These scripts execute only if any source video is available.

The correct behavior is "Add to avsi" → change processing parameters → "Make" → Go to a next step. In a coming beta version these steps are now simplier.

Also please wait the beta version. There are a lot of important bugs fixed. I.e. an incorrect color profiles used in import and export to H.264; a color shift in ut codec in RGB mode.

@rean Thanx a lot for your comments. I'll wait for your beta version. As you can see, your software is not that easy for a general user to understand, so a good manual is needed in any case.

Hi @rean,

Thanks for your beta version. I've already downloaded your beta and since you didn't release any tutorial, I would appreciate if you could confirm I'm doing everything right.

I'm on Win7 64bit. I use GH2 files. My goal: to get a video file transcoded to 4:4:4 16bit.

1. In order to get 4:4:4 I in "\wm 0.8.0 beta1 free\inc\" in settings.bat I changed "set wm_chroma_subsampling=422" to "set wm_chroma_subsampling=444".
2. Execute "\1. Src\Panasonic\Panasonic GH2.bat".
3. Execute "\2. Extract\Edit 01. Import.bat" (I didn't do anything here and closed the window).
4. Execute "\2. Extract\Edit All.bat" (I didn't do anything here and closed the window).
5. Execute "\2. Extract\Make.bat".
6. Execute "3. Edit\Avsi.bat".
7. Execute "3.Edit All.bat" (I didn't do anything here and closed the window).
8. Execute "\4. Make\Make h264 mov.bat"

According to these steps I got my file transcoded to Chroma Subsampling 4:4:4 and Bit depth: 8 bit!!! The new file is visually better than the original MTS and its size is much larger than the original. Why is it 8 bit and not 16 bit?

I also tried another export formats (Make DNxHD-10bit mov.bat, Make DNxHD-10bit mov.bat) and these transcoded files became: Chroma Subsampling 4:2:2 and Bit depth: 10 bit!!! The new file is visually the same as the original MTS and its size is almost the same as well. Why is it 10 bit and not 16 bit, and not 4:4:4?

I didn't use ProRes because I'm on a PC.

Am I doing something wrong?

Thank you in advance.

My workflow for my current project was DSLR footage -> Cineform and then edit within FCP7. How is the workflow with windmotion ? Do I need to denoise using Neat Video and then use windmotion ? So the workflow will be DSLR -> Cineform -> Neat Video -> Windmotion ?

Can you offer an output option to Cineform ? Or even to Prores ? I would be very keen on an output from 1080p or 720p to 2K 16:9 footage - with the 4:4:4 scaling applied.

Short Windmotion denoise "howto" (before I publish a full manual).

Usage: MotionDegrain8(clip c, clip m, int "bs", int "frames", int "th", int "limit")

Here are: c - original clip in the current channel (Y,U,V)

m - motion clip, used to restore details. These two parameters are default used for best quality. So, no any changes are required. Pro edition has a feature to tweak this motion clip to get better results in some cases.

bs - block size, used for motion calculation. Use 4 or 8 for detailed clip, 16 or 32 for very noise or less detailed clip.

frames - number of sequenced frames used to calculate motion and to denoise. Use values from 2 to 6. No any common requirements to choose, because it is a source kind-related. I use 2 to 4 in most cases. Bigger values creates motion artifacts (see below).

th - threshold to denoise. Use values from 1 to 10000. Bigger values enable more aggressive denoise, also you will get temporal artifacts (like a slow motion of fast movement objects). So better usage of this filter - to convert 50p(60p) to 25p(30p) or to denoise static scenes. Change a framerate after this filter. As a result you will get less strobe effect. I use 800 in my projects (50p->25p). You should keep this value minimal, for your source, to get less artifacts.

limit - frame change threshold limit in a brightness value range. In most case use 255(no limit). If you have artifacts, try to decrease it to 2 or 3.

Pro edition has another denoise filters, those create no temporal artifacts, or can better denoise of some clips.

I purchased the pro edition a couple weeks back, but never received any or of confirmation or download link. Was it just a pre-order for the eventual release of this beta version, or is there another step I need to go through?

has someone managed to use wm via wine or anything alike? i don't have a virtual machine set up, so that's not an option for me.

cheers, htf

@zcream I understand your request.

Probably in the future, but now I have no possibility to develop a special code for your codecs export. Windmotion supports only codecs from ffmpeg using conversion from a H.264 temporary file. (Why this strange way? I have removed DirectShow export possibility while development, because I found unknown bugs in third-party components. For this reason, two months ago, I decided to close the project. Also ffmpeg has no possibility to choose color matrix directly, so we have color shift in HD videos (known bt.709/bt.601 issue). But I found a conversion from H.264 file works fine, so I use this strange solution.)

Prores 4444 is not yet included in ffmpeg. I know this codec is included in ffmbc, but probably it is not supported for 10-bit conversion from H.264, so I do not support it. Also I found ffmbc is 3 times slow, without any superiority over ffmpeg.

Windmotion final release will support the same codec set as supported currently in beta. This includes: ProRes 4:2:2 10bit, DNxHD 1080p 4:2:2 10bit, H.264 lossless (all modes), Ut codec (all modes but RGB mode has color shift for HD source because a bug in ffmpeg), MJPEG 4:2:2, v210 4:2:2 10 bit.

I myself do not like the way the export is made currently. Currently this is a set of tricks to get accurate color over open source tools those developed by programmers not artists. But this is the best solution available today for me, which works fast and without errors. Perhaps in the future, I'll do something like ffmpeg, but for Windmotion especially, but not now.

@rean Could you post a small 1-2 second h.264 4:4:4 4K upscaled sample file ? I want to try run it via CF - convert to CF RAW and see if the FCP workflow actually works.

@rean, the Nex footage looks really good. Evident even with low quality youtube clip. How are you coming along with optimization for GH3 (based on sample-clips I sent)? When that is ready, I'll be buying the "Personal Edition" for sure! :)

i was eating peanut butter and honey !!! now i see this. Thanks @rean ,you make my day. Finally clear instructions. And thw video and image is awesom the before an after!! thanks!!

Here is the latest beta version. The next release will be the final release. I have successfully created a big 7-hour Full HD project with this version. The main documentation is also almost ready.

Version 0.9.0.Beta2, 2013-09-04

Audio processing disabled by default.
16:9 2K support 2048x1152.
Film grain filters are moved to different functions.
Film grain 16-bit export support.
Increased default crf for 720p film (Pro).
Script "Edit\Edit All Before After" without a histogram added.
Bugfix: Incorrect parameters of MakeFilm scripts fixed, setlocal added (Pro).
Bugfix: Private scripts removed from installer (Pro).
Bugfix: Incorrect mp4 mixing for Film (Pro).
x264 updated to the latest version (Komisar build is used).
TODO: Documentation in progress.

any idea on pricing?

From $0 to $275.

Thanks @humpman for the GF1 (hacked) footage 720p MOV MJPEG 120 mbps.

It is a resize example from 720p to 1080p with Pro edition sharp filters. Source image is pre-resized for 1080 for comparison. Denoise and dehalo filters and red and blue channel saturation are also used.

It is not very best resolution upsample example, because unfortunately the source video is blurry without subpixel details (probably GF1 lens is not so good for this fantastic hack bit rate). So I used existing 1-2 pixel sharp only. Subpixel sharp is useless here - I had only random dirty pixels.