v6.46+dev0131 - January 31 2022

[sockmonkey72]

v6.46+dev0131 - January 31 2022

• * Includes feature branch: batch converter improvements
• * Includes feature branch: improved zoomed-in peaks performance
• * Includes feature branch: FX chain oversampling
• * Includes feature branch: render internals simplification/cleanup
• * Includes feature branch: user-defined action overhaul
• * Includes feature branch: media explorer toolbar
• * Includes feature branch: media item fixed lanes
• + Batch converter: fix lost fx tail setting when closing/reopening dialog
• + Batch converter: save/restore dither, noise shaping settings with presets
• + Envelopes: do not select following envelope point when deleting via mouse click [t=262564]
• + Preferences: include mouseover help text in search terms
• # Batch converter: fix localization of list view column names
• # CPU metering: show separate menu items for per-core and total metering
• # Media explorer: fix configurable auto-crossfade length persistence

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

[Meo-Ada Mespotine]

I've done some tests with the extension of metadata-length using some testdata of 500MB.

It works good with GetSetProject_String, setting and getting with currently opened projects, so this works perfect.

However, I noticed, two other issues:

1) when I save a project with the user-metadata, it will only store the first ~8k of it and then truncate it.
It should be possible to store the entire user-metadata, imho.

2) when the user-metadata contains a \n in it, it will be replaced with a space. Would it be possible to retain \n in the project-file as well,
maybe storing it as BASE64, like multiline-strings in projextstates? Or does this have possible problematic implications with certain metadata-schemes?

FWIW, I generated the test-metadata with this code, using "ID3:TXXX:TUCH" for the test-metadata:

-- create 500MB test-metadata
A=""
B=""
C=""
-- kilobyte
for i=1, 1024 do A=A..string.char(math.random(255)) end

-- megabyte
for i=1, 1024 do B=B..A end

A=""

-- 500 megabyte
for i=1, 500 do C=C..B end

-- write and read metadata
A,B=reaper.GetSetProjectInfo_String(0, "RENDER_METADATA", "ID3:TXXX:TUCH|"..C, true)
X,Y=reaper.GetSetProjectInfo_String(0, "RENDER_METADATA", "ID3:TXXX:TUCH|", false)

-- check if strings are identical
F=C==Y

I've tested it on the 2901-dev-release, but I suppose, it's present in this release as well.

[Phazma]

Originally Posted by sockmonkey72

[*]+ Batch converter: save/restore dither, noise shaping settings with presets

I confirm that it works now.

[PhelixK]

• # CPU metering: show separate menu items for per-core and total metering

nice having this readout when not showing the graph area.

As for the option "show free system RAM rather than use", the readout remains the same here and doesn't report unused PhysMem. I'm wondering if it should do that on mac, and if it works for others?

[daxliniere]

Originally Posted by sockmonkey72

[*] * Includes feature branch: FX chain oversampling

Hey Justin, before this becomes mainstream, would you please be able to display the oversampling rate in a more visible place and also add some easier ways to set the OS, please?

For the first, I was thinking that the FX Chain window title could display FX: Track 1 [4x OS] and individual FX instances in the FX Chain window could display VST: SSL X-Saturator [8x OS].

As for setting OS rates for chains and FX instances, some actions would be excellent. Ideally;

Set FX Chain to no oversampling
Set FX Chain to 2x oversampling
Set FX Chain to 4x oversampling...

Set FX instance to no oversampling
Set FX instance to 2x oversampling
Set FX instance to 4x oversampling...

Increase/double FX Chain oversampling
Decrease/halve FX Chain oversampling
Increase/double FX instance oversampling
Decrease/halve FX instance oversampling

[Funkybot]

Originally Posted by sockmonkey72

[*]# CPU metering: show separate menu items for per-core and total metering

Please correct me if I'm wrong, but as I understand this new CPU calculation option, if I had a 32-track project with each track utilizing exactly 3% (with no variability) and a 32-core machine, Reaper would display the CPU as 96% of one core. What's the practical benefit or real-world use case for measuring CPU this way?

In that scenario, each core would likely be using 3% (ideal conditions) so nothing is anywhere near stressed. What useful information is Reaper conveying to me by showing me 96% in that scenario?

When I want to measure CPU usage, I generally want to know:

1. How much total headroom do I have on my CPU to keep adding plugins (current non-dev measure)?
2. What's the CPU measure on the most stressed core as this will cause glitches dropouts?

I don't think this new measure would bring someone closer to answering question #2 but it could confuse them into thinking it would. Unless I'm totally wrong on how it works.

[airon]

Old Protools has a nice meter that shows each core/threads use. Something like that ?

This article by Protools Expert explains it. The image below compares two different systems.



As for actions for controlling oversampling, if they're being considered, I'd expect to find them in the FX Chain window menus. Else they might get lost in the action sea .

[Phazma]

Originally Posted by airon

Aaaahh this looks clear! I always struggle finding out what's going on with Reaper's CPU meter. Something like this would make it easy to see at a glance how much CPU of the total CPU is being used, how much of which Core and also Disk Activity and available RAM.

The only thing not so clear is, if this is realtime or average CPU. If there was a button to toggle between the meters showing realtime vs average it would be perfect (or maybe even better if there were like 2 "channels" for each Core, one showing realtime and the other showing average.

[Loulou92]

Originally Posted by airon

Old Protools has a nice meter that shows each core/threads use. Something like that ?

This article by Protools Expert explains it. The image below compares two different systems.



As for actions for controlling oversampling, if they're being considered, I'd expect to find them in the FX Chain window menus. Else they might get lost in the action sea .

Would love such CPU metering.

For the oversampling action set, I invoke the famous meme : "why not both ?".

[BPBaker]

I get a repeatable crash on 6.46+dev0131 with MacOS 12.1 when opening the Batch Converter, then closing the Batch Converter window, then re-opening Batch Converter.

[Berg]

Originally Posted by BPBaker

I get a repeatable crash on 6.46+dev0131 with MacOS 12.1 when opening the Batch Converter, then closing the Batch Converter window, then re-opening Batch Converter.

I do too, same configuration.

[deeb]

Originally Posted by daxliniere

Hey Justin, before this becomes mainstream, would you please be able to display the oversampling rate in a more visible place and also add some easier ways to set the OS, please?

For the first, I was thinking that the FX Chain window title could display FX: Track 1 [4x OS] and individual FX instances in the FX Chain window could display VST: SSL X-Saturator [8x OS].

As for setting OS rates for chains and FX instances, some actions would be excellent. Ideally;

Set FX Chain to no oversampling
Set FX Chain to 2x oversampling
Set FX Chain to 4x oversampling...

Set FX instance to no oversampling
Set FX instance to 2x oversampling
Set FX instance to 4x oversampling...

Increase/double FX Chain oversampling
Decrease/halve FX Chain oversampling
Increase/double FX instance oversampling
Decrease/halve FX instance oversampling

I didn't try this oversampling features so i might be thinking wrong, anyway ..

Additionally to your suggestions how about:
"Set Oversampling to no as default for selected FX instance"
"Set Oversampling to 2X as default for selected FX instance"
"Set Oversampling to 4x as default for selected FX instance"


and then

"Set All FX for All tracks to Default Oversampling"


This way we could have easily both worlds: CPU less hungry and better latency and a switch to maximum quality when is appropriate in the project cycle.

[o_e]

Originally Posted by Berg

I do too, same configuration.

+1

Mac OS 10.14.6 aka Mojave

[AtmanActive]

I'm still hoping to see

Set FX Chain to 0.5x oversampling
Set FX Chain to 0.25x oversampling

Since I started mixing everything at 192kHz sampling rate, some fine old VSTs and VSTis are either crashing or keeping quiet. Would be great if I could run them at 96kHz or 48kHz while the whole project is still running at 192kHz.
Also useful for network streaming, like ReaStream, where the main mixing project is running at 192kHz while some remote monitoring can take place on an android phone locked to 48kHz.

Here is hoping.
Cheers!

[Hogi66]

Originally Posted by BPBaker

I get a repeatable crash on 6.46+dev0131 with MacOS 12.1 when opening the Batch Converter, then closing the Batch Converter window, then re-opening Batch Converter.

+1 on 6.46+dev0131 with MacOS 12.2

[Justin]

Originally Posted by Meo-Ada Mespotine

I've done some tests with the extension of metadata-length using some testdata of 500MB.

It works good with GetSetProject_String, setting and getting with currently opened projects, so this works perfect.

However, I noticed, two other issues:

1) when I save a project with the user-metadata, it will only store the first ~8k of it and then truncate it.
It should be possible to store the entire user-metadata, imho.

2) when the user-metadata contains a \n in it, it will be replaced with a space. Would it be possible to retain \n in the project-file as well,
maybe storing it as BASE64, like multiline-strings in projextstates? Or does this have possible problematic implications with certain metadata-schemes?

FWIW, I generated the test-metadata with this code, using "ID3:TXXX:TUCH" for the test-metadata:

-- create 500MB test-metadata
A=""
B=""
C=""
-- kilobyte
for i=1, 1024 do A=A..string.char(math.random(255)) end

-- megabyte
for i=1, 1024 do B=B..A end

A=""

-- 500 megabyte
for i=1, 500 do C=C..B end

-- write and read metadata
A,B=reaper.GetSetProjectInfo_String(0, "RENDER_METADATA", "ID3:TXXX:TUCH|"..C, true)
X,Y=reaper.GetSetProjectInfo_String(0, "RENDER_METADATA", "ID3:TXXX:TUCH|", false)

-- check if strings are identical
F=C==Y

I've tested it on the 2901-dev-release, but I suppose, it's present in this release as well.

Thanks, fixing the length issues now -- but be aware the metadata APIs are not binary-safe, e.g. a 0 byte in the middle of the string will terminate it (we may fix that at some point)