v6.53+dev0407 - April 7 2022

[sockmonkey72]

v6.53+dev0407 - April 7 2022

• * Includes feature branch: video decoding/encoding via Windows Media Foundation
• * Includes feature branch: internal pin management overhaul for future extension
• * Includes feature branch: display options for recent projects list
• * Includes feature branch: update ARA to version 2.1
• * Includes feature branch: insert new track when importing media via mouse drag-drop
• * Includes feature branch: render loudness graphing
• * Includes feature branch: media explorer user-defined tags
• * Includes feature branch: FX chain oversampling
• * Includes feature branch: media item fixed lanes
• + MIDI editor: fix editing left edge of notes beyond first loop boundary [t=264881]
• + MIDI editor: fix inserting new CC events past the loop boundary of looped items
• + MIDI editor: fix note velocity sweep-editing when moving the mouse quickly [t=264954]
• + MIDI editor: reload custom .ReaScale function when opening a new project
• + Media explorer: add a hidden accessible text field for pitch detection output
• + Preferences: include more information/help in video decoder details window
• + RF64: fix duplicate marker display [t=265078]
• + Razor edits: fix behavior when stretching right edge of razor edit with ripple edit enabled and automation items later on the track [p=2543398]
• + ReaControlMIDI: increase length of sysex dump displayed in log from 1k to 8k
  + Video list Windows Media Foundation details in available video decoder information window
• + Windows: support displaying shell metadata in .wma/.wmv media
• + Windows: support native decoding of .wma, .wmv, and mpeg-4 audio video via Windows Media Foundation
• + Windows: support native rendering of .wma, .wmv, and mpeg-4 audio and video via Windows Media Foundation
• # Color theme: more subtle default overlay for locked media items
• # macOS: improve appearance of TCP/MCP text in dark mode
  

This thread is for pre-release features discussion. Use the Feature Requests forum for other requests.

Changelog - Pre-Releases

Generated by X-Raym's REAPER ChangeLog to BBCode

[thommazk]

Originally Posted by sockmonkey72

+ MIDI editor: fix editing left edge of notes beyond first loop boundary [t=264881]

Thanks! The length editing is working now.
There's a related bug that's going on this thread with the 3rd gif:

The second gif is also happening but in a different way:

Originally Posted by sockmonkey72

+ MIDI editor: fix note velocity sweep-editing when moving the mouse quickly [t=264954]

Would you also check this related bug regarding moving the mouse quickly not erases midi notes? https://forum.cockos.com/showthread.php?t=263855

Thanks!

[schwa]

Originally Posted by thommazk

related bug

Fixing, thanks. The fast-moving-mouse-customized-right-click-erase-notes bug is more complicated, unfortunately.

[musicbynumbers]

Thanks for the new video support!

[Phazma]

Originally Posted by schwa

Fixing, thanks. The fast-moving-mouse-customized-right-click-erase-notes bug is more complicated, unfortunately.

I hope someday you will find some time to fix it, it is bothering me too. I also hope it won’t be too difficult to add the same behavior as mouse modifier option to the arrange view (for deleting items), after fixing it in the midi editor.

[gapalil001]

some thoughts about media item lanes

would be interested to try:

- make "media item lanes" button a part of TCP via WALTER

- Make an option that if a line is selected, the rest can be collapsed so that the selected lane is shown as the only and main one. Make the expand / collapse button of inactive lanes a part of WALTER (visible only if lane selected)



- make possible to allow script "SWS/BR: Toggle play from mouse cursor position and solo item and track under mouse for the duration" (and tons of similar) to play individual lanes or items even if they are not selected

- (more subjective one) do not allow "item_bg.png" to cross out of it's lane to avoid overlapping of item backgrounds that use yellow markers

[schwa]

Originally Posted by musicbynumbers

Thanks for the new video support!

Kind of excited about this tbh, in initial testing it's quite a bit faster to render video on windows natively vs using ffmpeg.

[Vagelis]

Originally Posted by Phazma

I hope someday you will find some time to fix it, it is bothering me too. I also hope it won’t be too difficult to add the same behavior as mouse modifier option to the arrange view (for deleting items), after fixing it in the midi editor.

+1

[Embass]

[Embass]

[BenK-msx]

Originally Posted by schwa

Kind of excited about this tbh, in initial testing it's quite a bit faster to render video on windows natively vs using ffmpeg.

Wild punt but would this development make recording video into reaper more achievable?
Cheers anyways of course.

[airon]

Originally Posted by schwa

Kind of excited about this tbh, in initial testing it's quite a bit faster to render video on windows natively vs using ffmpeg.

Thanks for putting in the effort here.

It seems to do at least as well on the first test I did with some ProRes video. I verified that the WMF was being used. Any performance uplift in this department is very welcome.

I track the seeking speed with the View: Move edit cursor to mouse cursor (no snapping) command. My repeat rate for the keyboard is as high as it can go and the delay until repeat happens is as low as it can go too. Makes for a great way to scrub video with a mouse. It's what I use all day.

Thank you.

[musicbynumbers]

Originally Posted by schwa

Kind of excited about this tbh, in initial testing it's quite a bit faster to render video on windows natively vs using ffmpeg.

Will give it a proper test when back Monday

One very sad thing for me though is that I noticed on first glance that it doesn't allow for pcm wav rendering for the audio part of the wrapper unlike ffmpeg does.

When I'm sending demo videos to clients for games. I tend to render using 16bit wav so the sound is uncompressed as I found there was always artefacts with at least the ffmpeg audio compression.

I'll give this one a go compressed and listen out for the classic "waterfall bubbles" in the high end. Be great if there was a wav option though

[PitchSlap]

+ Media explorer: add a hidden accessible text field for pitch detection output

Sorry for the can of worms, but how about sending it as MIDI if the ME output track contains a VSTi?
I don't think anyone will expect Melodyne but it could be useful.

FYI, maybe this thread from 11 years ago could be helpful for chords?

The 'secret' of polyphonic pitch detection revealed!
https://forum.cockos.com/showthread.php?t=74104

[schwa]

Originally Posted by musicbynumbers

Will give it a proper test when back Monday

One very sad thing for me though is that I noticed on first glance that it doesn't allow for pcm wav rendering for the audio part of the wrapper unlike ffmpeg does.

When I'm sending demo videos to clients for games. I tend to render using 16bit wav so the sound is uncompressed as I found there was always artefacts with at least the ffmpeg audio compression.

I'll give this one a go compressed and listen out for the classic "waterfall bubbles" in the high end. Be great if there was a wav option though

Unfortunately the only lossless codec the media foundation encoder supports is Windows Media Lossless, which is not ideal for interoperability. I expect REAPER will only support H.264 video with AAC audio using WMF, which tbf is probably the most commonly used combination.

[clepsydrae]

Originally Posted by schwa

Unfortunately the only lossless codec the media foundation encoder supports is Windows Media Lossless

Just spitballing here, and maybe this is annoying and uninformed, but: would some kind of automated remux-after-render solution be an option? Can of worms, maybe...

[MusoBob]

It looks right until you zoom in or play. It seems to be when you get further into the source file the timing goes out.
The section below goes out when zooming in or playing it starts in the source @ 2:14 other sections greater than that are out also.


Reaper-wma-Alignment-3.mp4

[schwa]

Originally Posted by MusoBob

It looks right until you zoom in or play. It seems to be when you get further into the source file the timing goes out.
The section below goes out when zooming in or playing it starts in the source @ 2:14 other sections greater than that are out also.

First off, what OS is this? The new wma/wmv support is Windows only, to be clear -- if you're not on Windows, sorry for the false hopes!

If you are on Windows, could you right-click the .wma items and show source properties, to confirm that they are actually being decoded with wmf, as opposed to ffmpeg or vlc?

If you are on Windows and the source properties do not say "using Windows Media Foundation decoder", can you open preferences/video and look at the video decoder priority there? If wmf is not listed first, move or add it so the list looks something like this:



(REAPER considers WMF a video decoder even though in this case it's decoding audio.)

[MusoBob]

On Win, the Win version uses wma and the Mac version uses m4a

Length: 3:12.040
Video: 0x0@1.#Jfps, Aspect=1.00, Decoding format=
Audio: 44100Hz 2ch 16bps
Using DirectShow decoder
Encoded audio format: WMAudioV9

Metadata:
Audio format: {00000162-0000-0010-8000-00AA00389B71}
Audio sample rate: 44100
Audio sample size: 16
Bit rate: 160016
Channels: 2
Creation software: Adobe Audition
Length: 1920410000
Protected: 0

[schwa]

Originally Posted by MusoBob

On Win, the Win version uses wma and the Mac version uses m4a

That says the wma file is being decoded using directshow, not windows media foundation. Can you check the decoder priority as shown in my previous post?

This is what you want to see:

[MusoBob]

I set it to OpenGL, restarted and same issue still using DirectShow

OpenGL - Automatic (• WMF (Windows Media Foundation), Windows Vista+ only, no additional download required)
wmf vlc ffmpeg qt:mov,qt,m4v,mp4 dshow:!mov,!mp4,!m4v,!qt

Length: 3:12.040
Video: 0x0@1.#Jfps, Aspect=1.00, Decoding format=
Audio: 44100Hz 2ch 16bps
Using DirectShow decoder
Encoded audio format: WMAudioV9

[schwa]

Thanks for taking the time to help out with this.

Please see my screenshot in post #18. It sounds like you are changing the "output" dropdown, but the part you want to change is the "video decoder priority" list. Please make sure the priority list has wmf listed first, as I have it in that screenshot, or at least before dshow.

After changing that you don't need to restart REAPER, but you do need to close and reopen the project.

[MusoBob]

I set it to
wmf ffmpeg vlc qt:mov,qt,m4v,mp4 dshow:!mov,!mp4,!m4v,!qt
but it's not using WMF
so I set it to force
wmf

and it's using WMFD but it's still out:

[Embass]

[Embass]

[bolgwrad]

Tried that last RE thing in free item positioning mode; it's OK until you expand the RE:

[schwa]

Originally Posted by MusoBob

it's using WMFD but it's still out:

Glad you finally got WMF to actually get called! Is it possible the audio timing is now correct but the peaks file is stale? What happens if you run the action “rebuild all peaks”?

[Embass]

1


2

[thommazk]

Originally Posted by schwa

Fixing, thanks. The fast-moving-mouse-customized-right-click-erase-notes bug is more complicated, unfortunately.

Thanks a lot! I hope you'll find the time to fix it at some point.
Is this moving-fast bug related to also not being able to delete really small MIDI notes when right-clicking and moving the mouse?

[schwa]

Originally Posted by musicbynumbers

One very sad thing for me though is that I noticed on first glance that it doesn't allow for pcm wav rendering for the audio part

Ah, I've just discovered that WMF supports embedding flac audio in mp4 although it's completely undocumented. We'll see if we can make that work too.

[Meo-Ada Mespotine]

Ohhhh, nice the mp4 audio/video rendering on Windows.

Is metadata also supported? Iirc, mp4 has ID3-support in general...

[Cinemascore]

Reporting that on my system (see below), any newer dev build than v6.53-dev404 gets me tons of audio cracklies. My current project has about 20 Omnispheres in it and all MIDI - perf load is about 13% so no CPU overloading going on.

So...all dev releases from dev404a, including today's dev407 all audio crackle for me.

System:

Windows 11 x64 (nconventional setup - nothing special)
Latest RME PCIe driver for ASIO audio.
Reaper dev builds as stated above (again, nothing special on its setup)

Any ideas? Last good build for me again was v6.53-dev404, Current produciton release is fine, too.

Please advise if you can. Thanks.

[Justin]

Originally Posted by Cinemascore

Reporting that on my system (see below), any newer dev build than v6.53-dev404 gets me tons of audio cracklies. My current project has about 20 Omnispheres in it and all MIDI - perf load is about 13% so no CPU overloading going on.

So...all dev releases from dev404a, including today's dev407 all audio crackle for me.

System:

Windows 11 x64 (nconventional setup - nothing special)
Latest RME PCIe driver for ASIO audio.
Reaper dev builds as stated above (again, nothing special on its setup)

Any ideas? Last good build for me again was v6.53-dev404, Current produciton release is fine, too.

Please advise if you can. Thanks.

If the issue is crackles due to CPU overrun, the important metric is not overall CPU use, but the RT CPU and RT longest-block values in REAPER's performance meter (you might have to right click in there to turn one of those on). Can you post comparisons of the problematic project's performance readings with +dev0404 vs +dev0404a?

[strachupl]

Originally Posted by gapalil001

- (more subjective one) do not allow "item_bg.png" to cross out of it's lane to avoid overlapping of item backgrounds that use yellow markers

This is also related to Midi Editor's Media Items Lane (MEMIL ). Maybe some option to display additional title bar area above items intead of yellow pixel* trick

[MusoBob]

Originally Posted by schwa

Glad you finally got WMF to actually get called! Is it possible the audio timing is now correct but the peaks file is stale? What happens if you run the action “rebuild all peaks”?

Rebuild don't work.
When it's zoomed out the peaks show as correct but play early.
As you can see when zoomed in the peaks moves, this is where it actually plays the hit from.
As said it only goes out after around the 2:14 mark in the source file, any source section after that time is out.
The other tacks are mostly using short source files for the instruments and don't seem to be out but the Drums have long files.
The whole file is inserted then the start an length adjusted with the script.
You can try this manually with the wma drum I uploaded
drag it in the set the start to 2:14.000
then do the same with the wav on the next track.

[schwa]

Originally Posted by MusoBob

Rebuild don't work.
When it's zoomed out the peaks show as correct but play early.
As you can see when zoomed in the peaks moves, this is where it actually plays the hit from.
As said it only goes out after around the 2:14 mark in the source file, any source section after that time is out.
The other tacks are mostly using short source files for the instruments and don't seem to be out but the Drums have long files.
The whole file is inserted then the start an length adjusted with the script.
You can try this manually with the wma drum I uploaded
drag it in the set the start to 2:14.000
then do the same with the wav on the next track

OK! I have come to the conclusion that this particular .wma file, the Drums_120.wma file you uploaded in this thread https://forum.cockos.com/showthread.php?t=264745, is in fact faulty.

Like most compressed formats, .wma embeds some sort of keyframe index so that the decoder can seek to a new position without having to decompress the entire file sequentially to get there. The decoder looks in the index for the keyframe immediately before the seek position, goes there, and then starts reading samples until it gets to the exactly correct timestamp.

The problem is that the index is not correct for this file. Specifically, after a certain point in the index, the keyframes start to point to a position that is *later* than the target position, so it's impossible for REAPER to recover by reading more samples.

In fact I can tell you exactly where the index fails, it's at 128 seconds (2:08) into the file.

To summarize, I think REAPER will play this file correctly, using the WMF decoder, if it's played from the start, or starting at any point prior to 2:08. If playback (or the zoomed-in display) starts at any point after 2:08, the decoding will be ahead of where it is supposed to be and the playback will be early and out of time.

For completeness, I'll note that the WMF decoder does seem to handle this file better than the ffmpeg decoder did. I think that the current +dev build of REAPER, using the WMF decoder, will convert this file properly using the batch converter, because the batch converter never needs to seek the file so the faulty index has no effect.

Please let me know if any of that isn't clear!

[MusoBob]

THANKS ! yep sure enough I tried another drum style wma and it was fine. So out of all the hundreds of drum wma's the one I chose had an error.
When I reset the codec priority it's now using the WMF but it would not do that originally until I just put wmf only in the list.

If someone else could drag this in their Win Reaper then right click source properties and see if it's using WMF.
RockHardLA%5e_120_Style.wma

[Cinemascore]

Originally Posted by Justin

If the issue is crackles due to CPU overrun, the important metric is not overall CPU use, but the RT CPU and RT longest-block values in REAPER's performance meter (you might have to right click in there to turn one of those on). Can you post comparisons of the problematic project's performance readings with +dev0404 vs +dev0404a?

Hi Justin,

Thanks for your reply.

It's not overrunning - the whole project performs perfectly as long is it's running on a build before dev-404a. To finish my cue work today, I just installed the 6.53 release version and it worked flawlessly.

I'll wait and see if this gets sorted out later. Thought I'd report in.

[ovnis]

New bug:

Issue with Peaks opacity. Now, it's only on/off.

[paaltio]

Originally Posted by schwa

Kind of excited about this tbh, in initial testing it's quite a bit faster to render video on windows natively vs using ffmpeg.

This is very cool, thanks! This makes me think: on macOS does REAPER utilize the hardware decode/encode support on Apple Silicon?