ComfyUI has been one of the most flexible, powerful, and efficient tools I’ve explored. Despite its steep learning curve, working on custom workflows and testing the latest AI models has been a valuable skill to develop.

In addition to this, learning how to efficiently prompt Black Forest Flux models, and how to train LoRAs has been a great combo.

On the paid side, Midjourney and Krea have been my go-to tools. Mastering both and mentoring teams on their use has made me an advocate for their potential—not just in generating high-quality visuals, but also in streamlining creativity. By integrating these tools effectively, teams have sped up ideation and enhanced overall productivity.

Below are four assets I created for a Midjourney Masterclass, along with a collection of test outputs from Krea’s real-time image generation tool and the ComfyUI Upscaler

For a personal project, I had the opportunity to use DiffusionBee. This Mac app has a simple yet effective approach to open-source generative models. It’s not very flexible, but it was reliable enough for what I needed.

Lately, I’ve been interested in the power of photo restoration. My workflow is simple yet effective:

1. Crop and straighten the image on Affinity Photo.

2. Use FLUX.1 Kontext for coloring and damage removal.

3. Split the image in ComfyUI with overlapping (2 columns, 3 rows for this image).

4. Upscale each part in ComfyUI using the same seed and low denoise for structural accuracy and higher resolution.

5. Stitch the parts back together in ComfyUI.

6. Subtle color correction on Affinity Photo.