2Dfy (DarkSky Edition) | Stylize 2.5D into 2D Anime (Better particles, light and contrast + HD Wallpaper training)

Important: Use the version of the LoRA depending on the resolution you are going to use. Safer weight: 0.2 or 0.3. Test until you find the right spot, from 0.2 to 0.75 is recommended. For some models you can also try 0.05. Stylize 2.5D or 2.25D generated images as 2D. You can use v1 for quality around 512-768 and HD_v1 for quality around 1024 (you still can use HD_v1 for lower qualities but it can make worse generations). Very recommended the use of face fix.

Trained LoRA stylized as 2D anime for the landscape/wallpaper resolution. It will work as a mini-model that will change the style of the model you'll use. For some models, it will be in a subtle way. The best use is in 2.25 to 2.5D models with good anatomy. You can use this LoRA to generate directly, to use it with the option "use as a base" (img2img in Stable Diffusion) or both, that will make the 2D style stronger. As I like images with good effects this LoRA will tend to do amazing results with effects.

It will hardly be influenced by the main model, so be careful which one will you use - as it will enhance it capabilities. Weight hardly depends on the model, I tend to use from 0,5 to 0,75. It can greatly help eyes and less for hands. Backgrounds and image will have the same style and sometimes background will be a little blurry because the focus will be on the character.

Tips: Try different weights for each model, start with 0.4 and slowly increase it until you find the right spot. Some models would need "semirealistic" in prompt to work, for others my LoRA won't work as great. Don't forget to use good negative prompts.

Some models would need "semirealistic" in prompt to work, for others my LoRA won't work as great.

This LoRA will change a lot your gen, it is not like other HD enhancers that will not change the composition. Remember, it has been trained with landscape/wallpaper size so it will be better that way, if you use portrait size it will still help but not as much. Tip: Create a good prompt, use a good model. If you have good results, test it with my LoRA.

V1.1+ to V2 changes: · New tags for anime names/character. Some bad images are being deleted, new ones for V2. · Recommended tags for the best quality: detailed, HD wallpaper, detailed eyes. (Also added new particles or effects like: glow, dust...) Custom style/pose/camera: blurry background, blurry vignette, painted anime style, movement effect, dynamic pose, dutch angle, medium view (between upper body and medium shot), full-medium shot. How to help to make good hands: Use "bad hands" in negative prompt.

This mini-model was trained with AnyLoRA so it should create good quality 2D anime. V1 is trained with half of my anime gallery, so that means 2271 images.

V1 was trained with resolution x640. It will work with resolution 768x512 / 832x448. HD_v1 was trained with resolution x1024. It uses resolution x1280 and x1024. (Since it has also been trained with panoramic images with certain aspect, in Stable Diffusion you can also test escalated resolutions from 950x512, like 1188x640) HD v1.1 was trained in resolution x896. You may notice some differences with v1.

Weight is very variable, but the best results I had is from 0,3 to 0,7, with 0,4 being the default for most models. You can also use from 0,75 to 1, but I don't recommend it. Weight will hardly depend on the model, test yourself. I like to use from 0,6 to 0,75 in my models mixed with 3D (2.5D). It can help the model to detail eyes with prompts like glowing eyes, detailed eyes.

Note: It was trained over SFW artwork so I don't know if it will work for N-SFW. If you have a lot of tags and you don't have good results, try a little more of CFG. If you want better results overall, try less complicated tags (so more generic ones), less "perfect" tags and try different negatives that can be good for the base model. If you use a lot of complicated tags, the combination can make worse the result.

Most used tags: solo (1821) 1girl (1791) long hair (1355) looking at viewer (1228) smile (800) dress (727) short hair (718) red eyes (715) bangs (692)

holding (608) weapon (601) black hair (566) blue eyes (550) ribbon (523) very long hair (500) gloves (496) skirt (495) sky (466) long sleeves (456) blonde hair (454) sword (418) flower (415) open mouth (401) shirt (401) upper body (397) hair ornament (397) bare shoulders (392) thighhighs (379) 1boy (379) white hair (372) hat (359) hair between eyes (356) blush (345) outdoors (344) petals (330)

bow (321) blue hair (309) twintails (306) cloud (306) cleavage (296) brown hair (296) holding weapon (292) wings (289) closed mouth (285) hair ribbon (283) male focus (270) yellow eyes (261) detached sleeves (253)

Now available to download in CivitAI https://civitai.com/models/279178?modelVersionId=314445

For more versions/editions check out CivitAI.