Project Overview
After all this time working with generative AI locally, I feel like I’ve finally reached the pipeline I was looking for.
When I started learning ComfyUI, I wanted to build a system that could produce portfolio quality work, run locally and fit into the same creative process I already use as a designer. Now that I’ve started working with local video generation, that milestone feels closer than ever.
One thing I often do in my personal projects is combine vector graphic design elements with 3D renders, photography or video. I had the idea of creating this character, but instead of just designing a final image, I thought what if I could pull back from that poster design and turn it into a full character video presentation?
With the idea of creating a final boss scene inspired by the Dark Souls games, I started building the project. Generative AI makes it possible to rearrange the usual order of production. I created the final visual first, treating it more like a poster, book cover or trading card design. From there, I generated and manually edited other image variations to fit the storyboard and build the sequence around that final composition.
Software
ComfyUI
Flux
LTX 2.3
Adobe After Effects
Adobe Photoshop
Adobe Illustrator
The key visuals were created and edited with Flux inside ComfyUI, a bit of ChatGPT and additional refinement in Photoshop. Once the styleframes were established, I used them for local video generations using LTX 2.3.
For this part of the process I used LTX Director, a custom node that allows much more precise control over timing and the flow of the generations. Being able to define when each action happens and how one styleframe transitions into the next made a huge difference. I can really recommend it.
The soundtrack was generated with Suno and then manually cut and rearranged to fit the pacing I wanted. The edit, timing, transitions, compositing and graphic design elements were all assembled manually in After Effects.
The project took roughly 15 hours to complete.Even though the assets were generated with AI, this was definitely not a “press a button and wait” process. The real work is still art direction, storytelling, choosing what matters, building a sequence, making decisions and shaping hundreds of small details into something coherent.
Copyright ©
All original artwork published by Enrique Ramos Boquera (circlenline) is licensed under “Attribution-NonCommercial-NoDerivatives 4.0 International”.
You must attribute the work to Enrique Ramos Boquera (circlenline) as the original author. Licensees may not use the work for commercial purposes unless they get permission.
Online presence
Unless otherwise specified or implied by the client, content (image, video or text) showing off any stage of a project may be featured on this website, social media or used in a portfolio/reel.