Preset Crawler Error: "Failure while crawling, couldn't get preset name"

[oliveroliverio]

Preset Crawler Error: "Failure while crawling, couldn't get preset name."

This error happens with some VSTs, and others (Serum, Sylenth, Refx Nexus) it get's the preset name fine. I also made sure to make the VST window floating and load only the VST2 version (not VST3 or AU).

If it's not too much to ask, I'd like to contribute to bug fixes like these and would like some guidance on troubleshooting and testing Realearn. I'm familiar with programming and could use more Rust programming experience.

[helgoboss]

Just seeing this now. Not sure if this is solvable. Some plug-ins don't report the preset name to the DAW. You can identify such plug-ins by checking whether the preset name appears in REAPER's own FX window header. Can you check if the ones that don't work report the preset name or not?

[helgoboss]

I mean it's probably not impossible ... one could do OCR to read the preset name, but that would be quite premium ;)

[oliveroliverio]

I see.... lol. I also see that you're using rust to write these extensions. I don't see very many OCR libraries for rust other than "rusty-tesseract" <https://github.com/thomasgruebl/rusty-tesseract?tab=readme-ov-file>. I have some coding experience, but not to the extent of a 9-5 software developer. What would you say the level of effort of this is? I could try tackling it myself as I like learning this stuff. Or I could donate a $100 if that's sufficient, possibly more if you make a video overview of how you would code this thing. Oliver

…

On Fri, Jun 21, 2024 at 6:52 AM Benjamin Klum ***@***.***> wrote: I mean it's probably not impossible ... one could do OCR to read the preset name, but that would be quite premium ;) — Reply to this email directly, view it on GitHub <#926 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABEFBTQXZRXFPOMNNODWC53ZIQV2HAVCNFSM6AAAAABH4VKJ2WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBSG44TQOJXGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[helgoboss]

This kind of thing can get ugly and challenging quite quickly, especially if you want it to work well. Even assuming that the OCR library is good enough for the purpose (which I wouldn't take for granted), this might involve quite a couple of additional challenges. At the top of my head:

• Capturing screen content as a bitmap that you can feed into the OCR algorithm, in a cross-platform way
• Acquiring the necessary permissions from the user (reading screen content is a sensitive topic)
• Taking care to not let the Helgobox distribution size get much bigger than it already is (who knows what amount of training data is necessary to make OCR work)

$100 ... in my country, this easily corresponds to 1 hour of senior freelancer work. No way to do this in 1 hour. So this feature would either need a lot of passion/motivation or substantially more funding or something in-between ;)

If you have the passion, it's best to try it yourself. Wanting something very bad can get you very far, in my experience (that's why ReaLearn exists ;) There might be libraries out there which help with above mentioned challenges. And if your only goal is that it works for your specific needs, not for others, it will lower the effort by much.