AniThing

CHECKPOINT
Original


Updated:

Recommended Settings:

  • Excessive negative prompt can makes your creations worse, so follow my suggestions below!

  • Before applying a LoRA to produce your favorite character, try it without first. You might be surprised what this model can do!

A1111 my settings:

I run my Home PC A1111 with this setting:

  • set COMMANDLINE_ARGS= --xformers

if you can't install xFormers (read below) use my Google Colab Setting:

  • set COMMANDLINE_ARGS= --disable-model-loading-ram-optimization --opt-sdp-no-mem-attention

My A1111 Version: v1.6.0-RC-28-ga0af2852  •  python: 3.10.6  •  torch: 2.0.1+cu118  •  xformers: 0.0.20  •  gradio: 3.41.2

If you want activate xformers optimization like my Home PC (How to install xFormers):

  • In A1111 click in "Setting Tab"

  • In the left coloumn, click in "Optimization"

  • in: "Cross attention optimization" select: "xformers"

  • Press in "Apply Settings"

  • Reboot your Stable Diffusion

If you can't install xFormers use SDP-ATTENTION, like my Google Colab:

  • In A1111 click in "Setting Tab"

  • In the left coloumn, click in "Optimization"

  • in: "Cross attention optimization" select: "sdp-no-mem - scaled dot product without memory efficient attention"

  • Press in "Apply Settings"

  • Reboot your Stable Diffusion

How to emulate the nvidia GPU follow this steps:

  • In A1111 click in "Setting Tab"

  • In the left coloumn, click in "Show all pages"

  • Search "Random number generator source"

  • Select the voice: "NV"

  • Press in "Apply Settings"

  • Reboot your Stable Diffusion

If you use my models, install the ADetailer extension for your A1111.

Navigate to the "Extensions" tab within Stable Diffusion.

  • Go to the "Install from URL" subsection.

  • Paste the following URL: https://github.com/Bing-su/adetailer

  • Click on the "Install" button to install the extension

  • Reboot your Stable Diffusion

How to install Euler Smea Dyn and Euler Max Sampler:

  • In A1111 click in "Extensions Tab"

  • click in "Install from URL"

  • Under "URL for extension's git repository" put this link: https://github.com/licyk/advanced_euler_sampler_extension

  • Once installed click in: "Installed" Tab

  • Click in "Apply and quit"

  • Reboot your Stable Diffusion

  • Now at the end of the list of the sampler, you have the new sampler.

How to use ADetailer with Euler Smea Dyn and Euler Max Sampler:

  • In A1111 click in "txt2img" tab

  • Expand and click in "enable ADetailer"

  • Scroll down and expand "inpaint" section

  • Click and turn on "Use separate Sampler"

  • Now select: "DPM++ 2M Karras" (or your favourite sampler)

NORMAL GENERATION SPOT that I found is:

  • Sampling method: Restart

  • CFG 7 -> Steps 25

  • Width: 576 (or 768)

  • Height: 1024

or:

  • Sampling method: DPM++ 2M SDE Karras

  • CFG 5 -> Steps 20

  • Width: 576 (or 768)

  • Height: 1024

or:

  • Sampling method: Euler A

  • CFG 5 -> Steps 20

  • Width: 576 (or 768)

  • Height: 1024

SUPERFAST GENERATION SPOT that I found is:

  • Sampling method: Euler (or Euler a)

  • CFG 2~3.5 -> Steps 10~15

  • Width: 576 (o 768)

  • Height: 1024

IN GENERAL FOLLOW THIS ADVICE:

As the CFG scale are raised, the steps must be raised. This is a kind of guideline:

CFG scale: 2~3 -> Steps: 10~15

CFG scale: 3~4 -> Steps: 15~20

CFG Scale: 4~5 -> Steps: 20~25

CFG Scale: 5~6 -> Steps: 25~30

CFG Scale: 6~7 -> Steps: 30~35

I didn't go any further, but it could (I say this theoretically) also work beyond 7, always increasing the steps step by step

As for the prompts... it's hard to help you here either.

They change based on what you want to do, whether realism, anime, cgi and so on.

I recommend you download an image from the main gallery and insert it into the PNG info tab to get the prompt used.

I also leave you, downloadable from here, my style.csv file with all the prompts I used. Just overwrite (if it is empty) the one in your Stable Diffusion folder (A1111)

MY FAVORITE PROMPT:

  • Realistim Prompt:

    (masterpiece, best quality, highres:1.2), (photorealistic:1.2), (film grain:1.3), (morbid light:1.2), (dynamic angle), raw photo, ->YOUR PROMPT<- , moonlight passing through hair, (random beautiful background:1.3), (sharp), exposure blend, bokeh, dim light, (hdr:1.4), high contrast, (muted colors, dim colors, soothing tones:1.3), low saturation, morbid

    Negative for Realism:

    cartoon, painting, illustration, (grayscale:1.4), normal quality:1.8, ugly, 3D rendering,

  • Hyper Realism Prompt (thanks to stapfschuh:):

    (((Ultra-HD-quality-details, real-photo-realistic, ideal))) ->YOUR PROMPT<-

    Negative for Hyper Realism

    3D-rendering:2, cartoonified:2, negative_hand-neg, bad-hands-5, (shadows on the face), (front light), ugly, fat, mole, moles, naked, nude, braless, cross, sepia, black&white, B&W, painting, drawing, illustration, octane

  • Anime Prompt:

    (#2D Anime Style), masterpiece, best quality, 2D anime, ->YOUR PROMPT<- dutch angle, cinematic, volumetric lighting, soft lighting, rim lighting,

    Negative for Anime:

    (worst quality, low quality), negative_hand-neg, bad-hands-5, realistic, photorealistic, photograph, NSFW, photorealism, realism, 3D, sepia, black&white, B&W,

  • Pixar Prompt:

    (masterpiece, best quality, highres:1.2), (Pixar CGI Style), (colorful), ->YOUR PROMPT<- Pixar-style, Character Design, CGI, 8k

    Negative for Pixar:

    (worst quality, low quality), negative_hand-neg, bad-hands-5, realistic, photorealistic, photograph, NSFW, photorealism, realism, 3D, sepia, black&white, B&W,

  • Disney Prompt:

    (#Disney CGI style), (masterpiece, best quality, highres:1.2), (intricate and beautiful:1.2), ->YOUR PROMPT<- (detailed light:1.2), (soft lighting, side lighting, reflected light), (colorful, dynamic angle)

    Negative for Disney:

    (worst quality, low quality), negative_hand-neg, bad-hands-5, sepia, black&white, B&W,

I like to use also these negative embedding:

Version Detail

SD 1.5
30
hope I can remember everything there is to write, because there are a lot of new things! Let's start immediately with the sampler used: Euler Max. (Infite thanks to nuaion https://civitai.com/user/nuaion who reported it to me) The more attentive of you will have noticed that I used this sampler and many will be wondering how to get it and activate it. Now I'll explain it to you straight away 1) Why Euler Max? - Because with 30 Steps (minimum) it gives surprising results. - Because it's fast - But above all because it was created to fix the hands problem on SD1.5. - It doesn't work miracles, but compared to before my hands are significantly improved! - Installing it is simple, since an extension has been made which you can find here: https://github.com/licyk/advanced_euler_sampler_extension (if you click on "show more" in the model description, I have included step-by-step instructions to install it , for those who are not familiar) 2) I have a new favorite Negative Embedding: CyberRealistic_Negative-neg.pt - https://civitai.com/models/77976/cyberrealistic-negative The reason? - Because it makes fewer composition errors than EasyNegative - With the same steps, sampler and prompt it is more realistic - Follows the prompt much better 3) Variety of faces - Finally there is more facial variety (not much but it is there) - Especially if: you deactivate the ADetailer (which I recommend doing if you intend to take images of: Close-up portraits and Anime in 2D version) - If you add a nationality next to the 1girl suffix (example: 1 venezuelan girl) - If you do not use negative embeddings, but use the suggested negative prompt: cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering 4) Improved the language in the prompt - It is no longer necessary to use (obviously if you want) words such as: masterpiece, best quality and so on - To create a realistic image just start the prompt with: "a photo of..." or what I prefer, "raw photo, your prompt" - For other styles, I always recommend starting the prompt with something like: (Anime Style), (Disney CGI Style), (Pixar CGI Style), (Pop-Art Style) and so on. For this type of style, I do not recommend using EasyNegative in the negative prompt because it will always tend to create "semi-realistic" or 2.5D images. So I would advise you to use the negative prompt without emebedding or with the cyerrealistic that I recommended a few lines above. I can't list all the prompts used, I recommend using the PNG info tab or looking at the image prompts in the galleries I created, with the styles you like. In broad terms these are the negatives used: For Realism: Negative prompt: (CyberRealistic_Negative-neg), cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian Negative prompt: Cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian Negative prompt: EasyNegative, cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian For Anime: Positive: ((Anime Style)), (2D), ((Cartoon)), YOUR PROMPT (perfectly-draw) Negative prompt: worst quality, bad quality, ugly, 3D, 2.5D, CGI, octane, realism, realistic, semi-realism, semi-realistic, (grayscale:1.4) -> Here, however, you can download the style.csv file (to be opened with Notepad++) where all my prompts are present I must also say that this version of AniThing (of which I am definitely proud) would not have been born thanks to a very light merge with the SG_161222 model: https://civitai.com/user/SG_161222. So really Thank you! Enjoy this version, experiment because there is so much to discover in my opinion and as always see you next week with the PRUNED version --------------------- ||VERSIONE ITALIANA|| --------------------- Spero di riuscire a ricordarmi tutto quello che c'è da scrivere, perchè di novità ce ne sono un sacco! Iniziamo subito con il sampler utilizzato: Euler Max . (Infite grazie a nuaion https://civitai.com/user/nuaion che me lo segnalato) I più attenti di voi avranno notato che ho utilizzato questo sampler e molti si chiederanno come averlo e attivarlo. Adesso ve lo spiego subito 1) Perchè Euler Max? - Perchè con 30 Steps (minimo) da risultati sorprendenti. - Perchè è veloce - Ma sopratutto perchè è nato per sistemare il problema delle mani su SD1.5. - Non fa miracoli, ma rispetto a prima le mani sono nettamente migliorate! - Per installarlo è semplice, visto che è stata fatta un estensione che trovate qui: https://github.com/licyk/advanced_euler_sampler_extension (se cliccate su "show more" nella descrizione del modello, ho inserito le istruzioni passo-passo per installarlo, per chi fosse poco pratico) 2) Ho un nuovo Embedding Negativo preferito: CyberRealistic_Negative-neg.pt - https://civitai.com/models/77976/cyberrealistic-negative Il motivo? - Perchè fa meno errori di composizione rispetto a EasyNegative - A parità di steps, sampler e prompt è più realistico - Segue decisamente meglio il prompt 3) Varietà delle facce - Finalmente c'è una maggiore varietà facciale (non molta ma c'è) - Soprattutto se: si disattiva l'ADetailer (cosa che consiglio di fare se si ha intenzione di fare immagini di: Ritratti ravvicinati e Anime in versione 2D) - Se si aggiunge una nazionalità vicino al suffisso 1girl (esempio: 1 venezuelan girl) - Se non si usano gli embedding negativi, ma si utilizza il prompt negativo suggerito: cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering 4) Migliorato il linguaggio nel prompt - Non serve più utilizzare (ovviamente se si vuole) parole come: masterpiece, best quality e via dicendo - Per creare un immagine realistica basterà iniziare il prompt con: "una foto di..." oppure quello che io preferisco, "raw photo, il tuo prompt" - Per gli altri stili, consiglio sempre di iniziare il prompt con tipo: (Anime Style), (Disney CGI Style), (Pixar CGI Style), (Pop-Art Style) e via dicendo. Per questo tipi di style, sconsiglio l'utilizzo di EasyNegative nel prompt negativo perchè tenderà sempre a fare immagini "semi-realistiche" o in 2.5D. Per cui vi consiglierei di utilizzare il prompt negativo senza emebedding o con il cyerrealistic che ho cosnigliato poche righe sopra. Non risco ad elencarvi tutti i prompt utilizzati, vi consiglio di utilizzare PNG info tab o guardare i prompt delle immagini delle gallerie che ho creato, con gli stili che vi piacciono. A Grandi linee questi sono i negativi utlizzati: Per il Realisimo: Negative prompt: (CyberRealistic_Negative-neg), cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian Negative prompt: Cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian Negative prompt: EasyNegative, cartoon, painting, illustration, (grayscale:1.4), (worst quality:2), (low quality:2), (normal quality:2), ugly, fat, 3D rendering, asian Per gli Anime: Positive: ((Anime Style)), (2D), ((Cartoon)), YOUR PROMPT (perfectly-draw) Negative prompt: worst quality, bad quality, ugly, 3D, 2.5D, CGI, octane, realism, realistic, semi-realism, semi-realistic, (grayscale:1.4) -> Qui comunque potete scaricare il file style.csv (da aprire con Notepad++) dove sono presenti tutti i miei prompts Devo anche dire che questa versione di AniThing (di cui sono decisamente orgoglioso) non sarebbe nata grazie ad un leggerissimo merge con il modello di SG_161222: https://civitai.com/user/SG_161222. So really Thank you! Godetevi questa versione, sperimentate perchè c'è tanto da scoprire secondo me e come sempre ci vediamo la prossima settimana con la versione PRUNED

Project Permissions

Reprinting is strictly prohibited

    Use Permissions

  • Use in TENSOR Online

  • As a online training base model on TENSOR

  • Use without crediting me

  • Share merges of this model

  • Use different permissions on merges

    Commercial Use

  • Sell generated contents

  • Use on generation services

  • Sell this model or merges

Related Posts

Describe the image you want to generate, then press Enter to send.