Well, the first beta of the Mac version of my Image Description Toolkit is available at http://www.theideaplace.net/projects/IDT-4.0.0Beta1Bld050.dmg.
-
Well, the first beta of the Mac version of my Image Description Toolkit is available at http://www.theideaplace.net/projects/IDT-4.0.0Beta1Bld050.dmg. This allows you to use local Ollama models or Claude or OpenAI models with your own API keys. I wrote a blog post earlier talking about the release but the Mac version wasn't ready. That blog post is at https://theideaplace.net/introducing-idt-4-0-beta-1-an-enhanced-way-to-describe-your-digital-images/. You need Ollama installed as I say but for me on an M1 Mac this is describing images at about 6 seconds per image with Moondream and 31 seconds with Llava.
-
Well, the first beta of the Mac version of my Image Description Toolkit is available at http://www.theideaplace.net/projects/IDT-4.0.0Beta1Bld050.dmg. This allows you to use local Ollama models or Claude or OpenAI models with your own API keys. I wrote a blog post earlier talking about the release but the Mac version wasn't ready. That blog post is at https://theideaplace.net/introducing-idt-4-0-beta-1-an-enhanced-way-to-describe-your-digital-images/. You need Ollama installed as I say but for me on an M1 Mac this is describing images at about 6 seconds per image with Moondream and 31 seconds with Llava.
A big, big warning if you use Claude or OpenAI models. I've done my best to ensure token use here is reasonable but given these services cost, please absolutely check your own account and the token use being logged by the app. I'll continue to refine my prompts and the way I handle images to maximize description quality and minimize cost but please do not start a large batch of images without testing one or two first.
-
A big, big warning if you use Claude or OpenAI models. I've done my best to ensure token use here is reasonable but given these services cost, please absolutely check your own account and the token use being logged by the app. I'll continue to refine my prompts and the way I handle images to maximize description quality and minimize cost but please do not start a large batch of images without testing one or two first.
The toolkit has two apps. ImageDescriber is a full UI experience for describing individual images or big batches. There is also a CMD tool, IDT, that you can use. There is a user guide linked with more details and I'll be creating more documentation and obviously refining the experiences. This is a beta and creating Mac software is new to me. The apps are both written in Python with AI assistance and at least here running well. I know famous last words. Issues at https://github.com/kellylford/Image-Description-Toolkit/issues
-
R relay@relay.mycrowd.ca shared this topicR relay@relay.infosec.exchange shared this topic