Red Tornado is a superhero from the DC Universe. This is an android that has been possessed by the spirit of air. He has the ability to create tornadoes that allow him to fly, he can also create storms to attack Lora made mostly using images from the 2009 comic and fan art. (Only one image from the original 1968 comic is used)Main trigger:<pre><code>red_tornado, 1man, bodysuit, cape
tornado_legs</code></pre>Additional options:<pre><code>black cape, empty eyes, joints</code></pre>May help:<pre><code>blank eyes tornado on legs, muscular, abs, wind </code></pre>"blank eyes" - for create no human eyes"tornado on legs" - the tornado_legs tag works very unstable. Sometimes you can substitute tornado on legs instead and you can get what you need. Or use these two tags together."abs" - for abswind - so that the cloak flutters in the wind or is raised upNotes:<ol><li>This lora can be used as a lora of the concept of tornado legs. I added to the selection several images of other superheroes who have tornado legs, as well as several generated images of girls with a tornado drawn below. But it works unstable. Be prepared to spend many attempts to get a good image.</li></ol><ol><li>Before training it was not clear how best to use the body tags. Probably stable diffusion would be better trained on the part of the body that it already knows as the upper body, and the tail from the tonado and its length would be designated with a separate tag. One way or another, the autotagger considered the tornado to be a part of the body and the following scheme was used during training:</li></ol><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/06df46cc-26be-4567-95dc-39e319822cc6/width=525/06df46cc-26be-4567-95dc-39e319822cc6.jpeg" />But in practice this scheme does not always work.

Red Tornado (DC Comics)

74 images versionThis lora on censor.art:<a target="_blank" rel="ugc" href="https://tensor.art/models/852179795574840883">https://tensor.art/models/852179795574840883</a>

Files with SHA256: 925535a85940b26c9f9a3b61b699463d005a5f92025498e541a9caa91b48d995