|

Light Image Resizer 7.2.1 × Ollama Soporta AI Vision

PorFabrice Horas

Light Image Editor AI Vision has been updated to support Ollama. You can now choose between using a cloud solution such as ChatGPT or Google Gemini API, or run one of the LLM vision models locally using Ollama.

Descargar Light Image Resizer
Download Ollama for Windows

What Is AI Vision?

AI Vision is the feature that describes an image and converts it into text.
You can use it on photos or screenshots for example, to:

  • Extract keywords and improve local search results
  • Add image descriptions to metadata
  • Detect specific content (e.g. images where people are smiling)
  • Combine description and analysis
  • Perform basic OCR-like extraction

What Is Ollama AI Vision?

Ollama is a program to run LLM models locally on your computer.
You can download models for free from the Ollama website and run them without an internet connection.

AI Vision in Light Image Editor

Light Image Editor is included in the package of Light Image Resizer. It is a Windows application to process single images—editing, upscaling, and now AI Vision.

With this feature, you can:

  • Add text descriptions directly into image metadata
  • Use custom prompts for analysis or keyword generation
  • Review previous prompts with the history button
  • Adapt the output depending on the image content and prompt

Why Is Ollama Support Important?

Compared to cloud APIs like ChatGPT or Google Gemini, Ollama offers two key advantages:

1. No API costs

Running the model on your own machine means no API credits or usage limits. You can analyze as many images as you want. It will be completely free as it’s an open-source project.

2. Privacy

Everything stays local—no need to upload private photos or documents. Ollama works offline, or over a VPN or Cloudflare Tunnel for remote access within your network.

Limitations of Ollama for AI Vision

Even if we’re here to help, using Ollama requires a few extra steps:

  • Install the Ollama software (easy)
  • Run a command line to download a vision model (see list here)
  • Requires basic technical skills (but we can assist)

Running advanced vision models also needs powerful hardware:

  • Minimum 16 GB RAM
  • A dedicated GPU is highly recommended
  • We tested Ollama with NVIDIA 4060 Super and 3090 (24 GB VRAM)

We can provide consultancy or remote help (Zoom, Google Meet, or Windows Quick Assist).

What Are the Best Models to Use with Ollama?

Model choice depends on your hardware, language, and use case.

Here are a few we recommend:

  • Gemma 4B – Fast and good language support
  • Mistral 3.1 – Slower but accurate
  • Qwen2.5-VL – New, promising, worth testing

Some models work mainly in English, while others may correctly support French, German, or Spanish.

See the dialog in Light Image Editor to choose your model:

Select your Ollama Model in settings. Gemma , MiniCPM, LLava, Qwen, Granite

What’s Next with Ollama and AI Vision?

Currently, AI Vision can run automatically when an image is opened in Light Image Editor.
It’s not a batch tool yet, but integration into Light Image Resizer may come based on user feedback.

This feature is also part of our #legaltechsoftware catalog of solutions, with use cases for:

  • Lawyers
  • Real estate professionals
  • Photographers

If you are a Photographer using Lightroom, we also recommend the project LR AI Assistant, specialized in adding keywords and descriptions to your Adobe Lightroom catalog.

Let us know if you want this feature to go further! As Obviousidea will integrate it in the Light Image Resizer batch part only if we have expressions of interest from our users.