Tag Archives: GenAI

#weeklyreview 08/2025

Posted on by
Reply

geez … I’m late with my review again….

Last week was a quiet one, with my friend Sam returning from her trip to Australia. We caught up and got back into our regular gym routine.

A modern gym with various fitness equipment, including kettlebells, dumbbells, and a barbell. Large mirrors reflect the space, showing a person standing in workout attire. Treadmills and weight racks are visible in the background. The lighting is bright, creating a spacious and inviting atmosphere.
Display screen of a Concept2 rowing machine showing workout stats: elapsed time of 30:09, stroke rate of 25 strokes per minute, current pace of 1:57 per 500 meters, distance of 6817 meters, average pace of 2:12.5 per 500 meters, split of 29 meters, and projected 14498 meters in 1 hour.



Our Sunday winter swimming sessions took place as usual, but this time we had to clear a patch of ice from the lake to get in the water. The ice was 5cm thick and needed to be cut through using axes. We managed to create a small path into the lake, which we used for our short dip. The water temperature was around 0.5-1°C, but it wasn’t too bad once we were in.

A partially frozen body of water with a large, dark patch of open water surrounded by thin ice and scattered ice chunks. Brown plants and branches are visible near the water's edge.
A person is emerging from a hole in the ice on a frozen lake, surrounded by snow. The sky is partly cloudy, and a forested area is visible in the background. There are reeds on both sides of the water opening.
A digital thermometer is partially submerged in a shallow, sandy water pool. The display shows two temperature readings. The water surface has visible reflections and bubbles.
Person in a swimsuit standing in partially frozen lake, surrounded by broken ice pieces, with a forest in the background and a partly cloudy sky.
A frozen lake with a snow-dusted shore and reeds on the right. The sky is partly cloudy with patches of blue visible. Dense forest lines the opposite side of the lake.


The rest of the week was spent experimenting with the Ollama tool. I’ve been testing various machine learning models and user interfaces, trying to get a better understanding of how they work. My goal is to set up a local installation of a LLM (Large Language Model) with Retrieval-Augmented Generation (RAG), which will allow me to securely work with my personal documents.

Thursday was again family dinner time and I made veggie wraps. Fried the fancy looking Enoki mushrooms. They are trick to eat without cutlery. Much too stringy and hard to chew off. Not recommended for wraps

A frying pan on a stove contains a colorful mix of chopped vegetables, including zucchini, red and green bell peppers, corn, and mushrooms. The ingredients are evenly distributed and lightly sautéed.
A hand holds a tortilla wrap filled with ingredients such as shredded cheese, guacamole, and diced chicken. The wrap is held upright, showing its contents from the top.
Two bundles of enoki mushrooms being pan-fried alongside sliced button mushrooms in a black frying pan.
A tortilla filled with guacamole, sliced chicken, red bell peppers, and topped with grated cheese, served on a plate with a fork and knife on either side.


Overall, it’s been a productive week, but mostly focused on testing and experimentation rather than any major achievements.

Upscale Videos using open source AI tools

Posted on by

I recently got the question from my uncle whether I can upscale one of his really old videos. The source was a short 10 seconds video with some low quality audio in 320×240 pixel resolution. Likely taken by one of the first video capable digital cameras or a phone many years ago.

I accepted the challenge as I had seen some AI tools like DiffusionBee being able to upscale images with decent quality.

I haven’t found a good free tool to upscale a video directly yet. There are shade free tools out there, but I don’t trust them.

What I ended up doing is exporting each frame of the original video to an image, scale up the images with an open AI model and then stitch them back together to a video.

1.) export each frame of the video to a JPEG file, export sound into a single file

ffmpeg -i input.mp4 ./LOW/frame_%04d.jpg
ffmpeg -i input_video.mpeg -vn ./output_audio.mp3

Directory Structure:

.
├── HIGH
│   ├── upscayl_jpg_realesrgan-x4plus_4x
│   │   ├── frame_0001.jpg
│   │   ├── frame_0002.jpg
│   │   ├── frame_0003.jpg
...
│   │   ├── frame_0254.jpg
│   │   └── frame_0255.jpg
│   ├── upscayl_jpg_remacri_3x
│   ├── upscayl_jpg_ultramix_balanced_3x
│   └── upscayl_jpg_ultrasharp_2x
└── LOW
│   ├── frame_0001.jpg
│   ├── frame_0002.jpg
...
│   ├── frame_0254.jpg
│   └── frame_0255.jpg

2.) Upscale images using AI tool Upscaly

https://upscayl.org/

brew install --cask upscayl
original image from video
upscaled 4x using the REALESGRAN model in upscayl

3.) combine new images into a movie

cd ./HIGH/upscayl_jpg_realesrgan-x4plus_4x

ffmpeg -framerate 15 -f image2 -pattern_type glob -i "frame_?.jpg" -i ../../output_audio.mp3 -c:v libx264 -crf 1 -vf scale=2048:2048 -pix_fmt yuv420p -vb 100M ../output_${PWD##/}.mp4