This week, let’s help Interior Designer Emma refresh a part of a house on a budget. Can you combine prompting and GenAI image generation to loop through various styles and produce visual proposals that Emma’s client will love?
Here is the challenge. Let’s use this thread to post our solutions to it, which should be uploaded to your public KNIME Hub spaces with tag JKISeason4-5 .
Need help with tags? To add tag JKISeason4-5 to your workflow, go to the description panel in KNIME Analytics Platform, click the pencil to edit it, and you will see the option for adding tags right there. Let us know if you have any problems!
And it includes all the objectives proposed in the challenge description.
A bit of flexibility implemented - the list of style is defined in a table and the room to focus on is defined in the String Widget.
Here’s my solution. Could not get a verification for OpenAI so I used Dall. This prevented editing the base image, but does produce credibly themed images.
I’ve just uploaded my solution for Week 5 of Just KNIME It!
In this version, I added a little extra — now Emma, our interior designer, can upload her interior photos directly and automatically get the results sent to her via email!
Super fun to build, and once again amazed by how flexible KNIME is, no heavy coding needed!
This challenge really made my API usage at https://platform.openai.com go up like crazy
But that’s okay — if it helps Emma get the solution she needs, it’s totally worth it!
Fingers crossed she enjoys using it… and that my API quota survives this season!
Wondering the same. In other workflows I have already experimented with HuggingFace models. However, I do not seem to be able to get it work with generating images… Curious whether a solution exists
Unfortunately, I am not aware of good free and open-source models for image generation and editing.
Some of the vision models that you find on Hugging Face, for example those provided by Stable Diffusion (https://huggingface.co/stabilityai/stable-diffusion-3.5-large), are good but still require some form of pay subscription if you want to use them for out-of-the-box API inference, or some quite technical fiddling if you want to use the files locally.
On Ollama, you can find some decent vision models, for example LLama 3.2-vision (https://ollama.com/library/llama3.2-vision), but this model is designed for visual recognition, image reasoning, captioning, and answering general questions about an image (see an example workflow with Llava model). So, not really image generation and editing.
But please let me know if you find a good free model for image generation and editing. Even better if it’s available via a an easy to use and free API .
For image generation it’s trickier, I used python view node, where I called a huggingface model which generated the image for me. As it is more complicated I created an example workflow for that:
Hi Roberto! I just see your reply now (after replied to Soybeans0000023)
I think on Huggingface you can find free models to generate images for you, but as KNIME doesn’t have built in image nodes for Huggingface AI you have to call it from python (with a python view node).
I have created a workflow how I created images with python view (uploaded to my public Hub):
For that example, I used the “black-forest-labs/FLUX.1-schnell” Huggingface model: