A practical image-to-image workflow for Strike Fighters 2 skins

[jamieklausen998]

I have been revisiting how I build custom skins for Strike Fighters 2, and this may help others who enjoy the texture and modding side of the series. An image-to-image workflow is a useful middle ground between painting every detail by hand and accepting a fully generated image. I treat the result as a starting layer, not a finished skin.

 

I start with a copy of the original aircraft texture and a clean reference sheet. I keep panel lines, markings, rivets, and camouflage on separate layers. Then I use a controlled pass to explore variations in color, weathering, fabric texture, or a period-appropriate pattern. Low transformation strength is more useful than a dramatic re-imagination because it keeps the fuselage and wings recognizable.

 

Next I clean the result manually. I check that wing roots line up, serial numbers are readable, and mirrored parts are consistent. I also restore transparent areas and alpha channels rather than assuming the generated file preserved them. For metal aircraft, I compare highlights against the model in the sim; soft highlights can make an airframe look like plastic, while too much noise can make a simple finish look artificial.

 

Before packaging, I test the skin in different lighting conditions and at normal gameplay distance. A texture that looks good zoomed in may be too busy when the aircraft is moving quickly. I keep a layered source file, a flattened game-ready texture, and a short note describing what changed. This makes corrections easier and shows which parts were hand-painted or assisted.

 

For reference-guided experiments, I compare settings and do the final cleanup locally. This approach is best used to generate options for a skin concept, not as a replacement for checking UV alignment, game conventions, or the original author's work. Has anyone else tried this for SF2 aircraft or cockpit panels? I would be interested in the settings and cleanup steps that work best. My current comparison point is image to image ai