LinuxMusicians

JamesPeters wrote:And by "later" I mean I just tested it (Linux VST in Reaper).

CPU is back to the same levels as in 1.1.4, yay! Drawing of the graph is fast too.

Nice!

JamesPeters wrote:Yes I noticed that the late reflections of Room are significantly quieter. It didn't bother me, but if others are mentioning it, it might be worth considering doing something about it.

Yeah, adding some gain would be really easy, it's just a matter of figuring out how many decibels we want, and then multiplying all of the numbers that come out of the late dsp by the corresponding coefficient.

Sigh... These plugins are having a crisis of identity. Some people report that their favorite room preset is the "clear room" of Dragonfly Hall, and other people say that they really like making a hall sound with the Dragonfly Room algorithm.

I named them based on my perception of their utility, but with so many opinions I wonder if it would be better to give them more generic names, maybe something like "green dragonfly" for the original one based on hibiki and "blue dragonfly" for the one based on ProG (along with a corresponding change in color scheme for the new one).

That works for me.

In Renoise I get now ~+3 % DSP Load (Jack) with Dragonfly Reverb and +9% with Room.
In Reaper I get +0% for Reverb and +0.1% for Room.


@James: What are your numbers now? Strange that it's much worse in Renoise.

My numbers in Reaper (the only DAW I use) are back to what they were for Dragonfly Hall 1.1.4, which is approximately 0.7% CPU on my desktop Linux computer. (0.8% for hall, 0.6% for room with the oversampling at 1x.) The graphs draw very quickly too.

This is for the Linux VST versions of the Dragonfly plugins, version 1.9.3.

Performance is similar in Windows for the Windows VST version 1.9.3, but my laptop (which still has Windows) has a less powerful CPU than my desktop, so I'm doing relative comparisons (version 1.9.3 to 1.1.4). It's the same CPU usage for those versions. So I'd say the CPU usage issue seems resolved at least for Reaper, using VST.

Beta release 1.9.4 adds three new dials to Dragonfly Room: Input LPF, Output LPF, and Boost Band (frequency bandwidth of the bass boost feature).

Beta release 1.9.5 is hot off the press. Dragonfly Room's Output LPF previously only applied to the late output, now it applies to both early and late. Also, the following parameters have been exposed to the DAW (They aren't visible in the GUI, as they are considered more advanced features, and I don't think the GUI needs any more complication):

• Early Send (Dragonfly Hall and Dragonfly Room)
• Spin Factor (Dragonfly Hall)
• LFO Factor (Dragonfly Hall)

The latter two will hopefully be useful to increase or decrease the "tape flutter" effect that some people have noticed in Dragonfly Hall.

lilith wrote:In Renoise I get now ~+3 % DSP Load (Jack) with Dragonfly Reverb and +9% with Room.
In Reaper I get +0% for Reverb and +0.1% for Room.


@James: What are your numbers now? Strange that it's much worse in Renoise.

That's most likely because of reaper's "anticipative fx processing". Reaper normally precalculates the dsp for all tracks that aren't processing live sound. To get a value that you can compare you'd have to recarm the track or place the fx on the masterbus in reaper.

Jack Winter wrote: That's most likely because of reaper's "anticipative fx processing". Reaper normally precalculates the dsp for all tracks that aren't processing live sound. To get a value that you can compare you'd have to recarm the track or place the fx on the masterbus in reaper.

I just tried them on the master in my Reaper test project, and the numbers I got there for CPU are comparable to the CPU when the plugins are used on tracks with anticipative processing (actually they use a little bit less CPU on the master). Same for when I record-arm a track with one of these plugins. Except that the RT CPU doubles when I do this (in either case, on the master or a record-armed track). On my system it means going from maybe 1.5% RT CPU to maybe 3% RT CPU (done for a single instance of either plugin).

The RT CPU measurement has little to do with the CPU measurement, RT CPU is similar to JACK's DSP load, that is to say that it measures how much of the time available to process the audio was actually used.

Jack Winter wrote:The RT CPU measurement has little to do with the CPU measurement, RT CPU is similar to JACK's DSP load, that is to say that it measures how much of the time available to process the audio was actually used.

Ok then. That means the "true CPU" of these plugins is 15% lower when used without anticipative fx processing (and/or on the master bus) than the figures I showed earlier. If you ignore how the RT CPU increases.

Beta release 1.9.6 is here.

This one includes several bug fixes, a change in the damper/lpf dials, and a new look for Dragonfly Room:

Beta release 1.9.7 is available. The biggest change is that the first knob is now "Size" instead of "Oversample", which had a variety of issues.

@Michael Willis

Really enjoying both 'Verbs they cover a lot of ground (or perhaps space) thanks for your work!

GMaq wrote:@Michael Willis

Really enjoying both 'Verbs they cover a lot of ground (or perhaps space) thanks for your work!

Aww, thanks! I'm hoping to have Dragonfly Room stabilized soon; it's so hard to make up my mind on some of the minutia. Once I'm done fiddling around with the dials, I'll cut a proper 2.0.0 release, and then I will make sure that subsequent versions remain backward compatible.