This is my attempt at training a LoRA for the comic/animated version of John Constantine.<hr />V2 (clip skip 1) is my second attempt and has much improved tagging done on the training in addition to adding details to account for revealing/partially clothed and NSFW prompts.V2 main tags:<pre><code>john_c, 1boy, blonde hair, light blue eyes, thick stubble </code></pre>V2 outfit tags (2 tagged necktie colors):<pre><code>tan trench coat, white collared shirt, red necktie, belt, dark brown pants, socks, black shoes</code></pre><pre><code>tan trench coat, white collared shirt, black necktie, belt, dark brown pants, socks, black shoes</code></pre>All of my testing was done on Forge with DPM++ 2M SDM - Karras - (CFG 3-5) and this LoRA was trained to be clip skip 2 and I use an adetailer.<hr />V1 (clip skip 2) was my first attempt and was fully trained using screen grabs taken of his animated version.V1 main tags:<pre><code>john_c, 1boy, blonde hair, light blue eyes, stubble </code></pre>V1 outfit tags:<pre><code>tan trench coat, white collared shirt, necktie, dark brown pants, socks, black shoes</code></pre>All of my testing was done on Forge with DPM++ 2M SDM - Karras - (CFG 5-8) and this LoRA was trained to be clip skip 2 and I use an adetailer.

Files with SHA256: 4db6f7731e0552ac162cc1dd345be82b5c7cd25cb622f91bbfe529f0c3cb09e2