@FreakyFwoof @jcsteh I know others did too, but two features I kept raising with the NI team over many years were automatic parameter mapping and the ability to control non NI effects. It looks like they are finally here! :)

@FreakyFwoof Wow, the level of integration with Logic is awesome! I suspect the integration with other DAWs is not as comprehensive?

#InspiredBySound - Native Instruments Kontrol MK3 Accessibility Demonstration https://youtu.be/cMkm51Utijs

#Ollama v0.14.1 has Experimental image generation models.
ollama run x/z-image-turbo
Only available in Mac Silicon and Linux with Cuda, and apparently more models are coming soon such as GLM-Image, Qwen-Image-2512, Qwen-Image-Edit-2511... #LLM #ML #AI https://github.com/ollama/ollama/releases/

@vick21 @ZBennoui @BorrisInABox @Jage @ToniBarth That's cool. Certainly Gemini-cli uses routing strategy to decide which model to use in the beginning, but I'm not sure it keeps evaluating on the fly which model to use as things progress.

@miki @FreakyFwoof Yeah or even simple docker or VM would work.

Fun fact: I routed Claude Code to a local model and captured its system prompt. It is about 16K tokens long, compared with 6.5K tokens for Codex and 5.5K tokens for Gemini CLI. Some of the length is due to tool call descriptions, but I didn't have any additional MCP. No wonder why you can't effectively use local models with it unless you have a crazy machine that can run a bigger model capable of following lengthy instructions and handle very big context window. #LLM #AI https://gist.githubusercontent.com/chigkim/1f37bb2be98d97c952fd79cbb3efb1c6/raw/02b63535918082d98797b152f1513075d8aef7ab/claude-code.txt

@miki @FreakyFwoof It jus saves a little steps. Open editor, write, select all, copy, paste. With external editor, you can just press control+g write stuff, and save.

@FreakyFwoof BTW, one of the features I use often is to compose a long prompt in an external editor by pressing ctrl+g. For example, I set my EDITOR to notepad++ via environment variable. setx EDITOR C:\Portable\Notepad++Portable\App\Notepad++64\notepad++.exe

@FreakyFwoof Also you might want to try dangerous full auto mode where it doesn't keep asking you to confirm before doing something. Like the name says it's dangerous because you have no idea what it would do beforehand. lol

@FreakyFwoof Now have fun! I was able to make a simple wave editor by simply describing various workflow that I like from different DAWs I use. lol

@FreakyFwoof You can just get the installer from the github releases. https://github.com/microsoft/terminal/releases

@FreakyFwoof For my curiosity, can you let me know when you resolve the issue and how?

@FreakyFwoof Yes you should definitely try the new Windows terminal. It's so much better. I.E. Select all with ctrl+shift+a, clear screen with ctrl+shift+k, and so many more shortcuts and customization. It replaces powershell, command line, wsl, etc. It even has screen reader support.

@FreakyFwoof Oh also I'm using the new improved Windows terminal. I wonder it has to do with that? https://aka.ms/terminal

@FreakyFwoof Yes I used npm i -g @openai/codex

@FreakyFwoof That's very weird. I just blow out codex settings folder, uninstalled and installed the latest, and I was able to login and use it in windows commandline not powershell.

@FreakyFwoof That's very odd. It has some weird characters for TUI, and I wonder it has to do with synth you are using.

@FreakyFwoof Hmm, that's interesting. I often use Codex with NVDA, but never crashed. Not related, but are you using Codex with a local model or ChatGPT Plus sub?

@ZBennoui Pretty cool. I have no idea what's going on under the hood, but I assume it morphs the notes and tempo of different recorded audio loops to adopt to your project?