Ollama error ResponseChat when giving to large image to a vision model

[ZaMeR12]

Hi, I had some problems to send some images to my visions model to my ollama api and caused some error. Until i found that is the resolution of the image that make this error. I mean, all my image of png and jpg work well excepted for one who had bigger resolution(the one who caused the error). So i ask if there is a way to know the max resolution of images that i can give to ollama? Because i don,t want to guess the resolution with hundred tests.

[hopperelec]

I've not actually used any vision models before, but my understanding is that the max image resolution depends on the model you use, and a quick Google search should show you what that is for the model you are using

[ZaMeR12]

I've not actually used any vision models before, but my understanding is that the max image resolution depends on the model you use, and a quick Google search should show you what that is for the model you are using

Thanks for the answer. The problem is i did my research and even like in hugging face (library for majority of ai models), there is no clearly prefered size of image for request. There is more the memory of the context, but to calculated an aproximative ideal image resolution with the context's memory is not something i want to do for each model. So if that was really the case of a standard that dev who generated the model, tell about a recommended resolution for a request. Yeah that will be easy to do a "standard" max recommended image resolution for all my models. But the truth is that the dev doesn't do that for x or y reason.

[hopperelec]

Most AI models are based on some core model such as Llama 3.1 and it is this core model where these kinds of constraints would come from. Would you mind sharing which model you're using and I could have a look for you?