<a target="_blank" rel="ugc" href="https://ko-fi.com/lilllithhhhh">Support me here via my Ko-Fi &lt;3</a>Lighting is the hardest thing about AI image generation by far. Even if you specify the lighting direction with keywords, the level of control we have for lighting is primitive. It is very hard to get the desired lighting that you actually want, not just generic lighting setups like from the left, right, etc, but actual complex lighting setups that are hard to specify in just words.Hobbyists probably don't care much about lighting, but people who are serious with getting the right results require it, because currently lighting is for all intents and purposes completely random. You cannot specify the distance the AI generates the light (eg. 40ft away), nor the specific X, Y, Z angle of the light. There have been other attempts to light up a scene, such as <a target="_blank" rel="ugc" href="https://civarchive.com/models/2305167/qwen-edit-2511lightingremapalpha02">Lighting Remap Alpha0.2</a>, which help position lights in various positions. This does help, however, it still fails to capture the exact lighting position since it's based on a 2d canvas.Wouldn't it be better to have some sort of semi real-time lighting setup where you can roughly make sure the character is lit up the way that you want it, while not being bound to two-dimensions? Well, with a mix of a couple LoRAs (<a target="_blank" rel="ugc" href="https://civarchive.com/models/2346854/to-material-preview-style-or-qwen-edit-2511">Material Preview style</a>), <a target="_blank" rel="ugc" href="https://huggingface.co/microsoft/TRELLIS.2-4B">Trellis2</a> model generation (or Hunyuan 2.1/3.1), and <a target="_blank" rel="ugc" href="https://www.blender.org/">Blender</a>, we can have more precise control over the lighting in the scene.<h2 id="how-do-you-use-it">How do you use it?</h2>In order to use it, you must:1). Have an image preprocessed with my <a target="_blank" rel="ugc" href="https://civarchive.com/models/2346854/to-material-preview-style-or-qwen-edit-2511">Material Preview style</a> LoRA. (Image 1)and2). Have a Blender render (Cycles or Eeeve, doesn't matter) of a Trellis2 or Hunyuan 3D model of the character in the same pose. Light up your scene the way you fancy it in the Blender viewport. (Image 2)3.) Once you have those two images, put them into Qwen Image Edit 2511 and only use the AnyLight LoRA with the following prompt:<pre><code>Apply the lighting from image 2 to image 1.</code></pre>Use at strength 1. While you can use this with the Lightx2v 4-Step Lighting LoRA, it is recommended to use the 8-step Lightning LoRA for better results. Also, bumping up the resolution to 1080p additionally improves the results, so use that for optimal performance.NOTE: Blender renders of the same scene don't have to be exact, i.e. the models don't have to match up perfectly (angle, spacing, etc of the character). However, the Blender camera should generally reside in the same place as the input image. Getting as close to the exact angle and positioning in the input image via the Blender camera gives much better results. Additinally, this model is trained through the use Material Preview style renders, meaning that the use of other images in other styles will incur style drift. Also, this LoRA is geared towards pinup-like compositions.

Files with SHA256: 1417a0d7c1a1e28eee877fcad37751b5e9a4749fd13de1d3fe3002635a52cd28