2DN-LatentLens

2DN-LatentLens v1.0

Update: you can play around with the model for free using my Discord bot https://discord.gg/3t2CAF9DRj

It uses the right sampler and scheduler settings not available on CivitAI or any other online image generator service. It is also very fast.

Introducing 2DN LatentLens.

LatentLens is a v-pred model, based on NoobAI XL v-pred v1.0

It uses a novel technique called LatentLens, which increases model responsiveness to prompts, coherence and quality. For example, most of the preview images use neither hires fix, face detailers nor upscaling. LatenLens makes models more flexible, allowing you to explore various creative prompting strategies while seeing consistently coherent and vibrant generation results.

This model works with NoobAI artist tags (not pictured) and is good at explicit content (not pictured).

My intention when working on this model was to "let people's dreams come true easily".

For best results:

Include in positive prompt:

masterpiece, best quality, amazing quality, very aesthetic, absurdres, very awa,

Include in negative prompt:

lowres, worst quality, low quality, bad anatomy, bad hands, multiple views, abstract, signature, furry, anthro, 2koma, 4koma, comic, manga, text, logo, artist signature, patreon logo, patreon username, twitter username

Recommended settings:

Guidance: 1.1

Steps: 20

Sampler: DPM++ SDE CFG++

Scheduler: SGM Uniform

Software: reForge

The model is also available on TensorArt

2DN-Illust v1.0

Introducing 2DN Illustrious.

This model uses IllustriousXL/NoobAI as a base. It was fine-tuned using 65,000 high quality photos for 12 epochs, two of which were preconditioning. Note that I specifically avoided training on cosplay photos and selected a very diverse dataset (LAION etc). It was further refined with custom LORAs (such as my cfg scale LoRA).

For those of you coming from Pony there are some important things to note:

• Your Pony LoRAs won't work. This is not as big of an issue as you might imagine. Illustrious/Noob has more than 100x the concepts and tags available for use than Pony. Including obscure tags you can only find on Danbooru. It also has tens of thousands of artist and characters tags if you want to make anime images or mix them into the realism style.

• You don't need to use the pony score tags. They will only interfere with the images.

REALISTIC MODE

I recommend the following base prompt:

highly detailed, realistic, fine fabric detail, absurdres, highly-detailed, best quality, masterpiece, very aesthetic, <your prompt>

For the negative prompt you should almost always use:

lowres, worst quality, low quality, bad anatomy, bad hands, multiple views, abstract, signature, furry, anthro, bkub, 2koma, 4koma, comic, manga, sketch, ixy,

Inserting artist tags of creators who make highly non-realistic artworks such as ixy and bkub improves the quality of the outputs.

Using artist names in the positive prompt will make the model produce anime style images. For a 2.5D look reduce the weight of the artist tags to 0.5 for example.

Facial expression tags will result in exaggerated results. Lower the weight of these tags also to control their effect when generating realistic images.

Lower CFG scale settings will produce less saturated images with lower contrast.

You can control the temperature/colour cast of the images by specifying for example „warm light” in the positive prompt and „blue, blue light” in the negative prompt.

If the model produces „babyface” during the ADetailer stage, set ADetailer positive prompt to an anime artist name such as „quasarcake”.

ANIME MODE

Simply prompt as you would any illustrious model, including artist tags.

—

Highres-fix or upscaling is highly recommended (0.3 denoising strength works well). ADetailer is essential for non close-ups of subjects.

Recommended CFG scale is 6.5. Adjust to taste.

Training was performed on a RunPod A100 GPU, costing $200USD. The model will be buzz gated for a short while. Your money will go towards offsetting the training costs.

2DN-Pony v2.0

I'm happy to release v2.0 of the original 2DN-Pony series! The new model has many improvements across the board:

• Better anatomy

• More pleasing and focused composition

• Finer details

• Wider dynamic range

• Deep tonality the original model was known for

• Works with a bigger range of CFG scales

It is being released for a short time under the EA program to help support further development.

Note the model is highly detailed, so you do not need to use detail/slider LoRAs to get fine structure in your images by default.

2DN-Pony v1.0 NoST

As thanks for those who have donated buzz for early access PLaY I’m releasing NoST(algia). This 2DN version aims to replicate the feel of SD1.5 models, including ridiculous detail, sameface 😂, rich backgrounds, and too many mechs when you use the mecha token. Unfortunately (?) it’s surprisingly good at hands and fingers. Motivated in part because I looked back on my old SD1.5 gens and there was something fun and kooky about them that got lost along the way.

Announcing 2DN-Pony v1.0 PLaY

Update: site gen isn't really working for some reason, sorry!

Merging reality with fantasy... Let your dreams run wild!

My best model and biggest fine-tuning project to date.

Trained on 70,000 selected photos.

I used the buzz donated from 2DN ReAL and put it toward cloud GPU compute to fine-tune this model. Any donations on this model will 100% go towards training new models (after CivitAI's cut).

Good for SFW, and NSFW. Pony Lora compatible but keep in mind this is a realistic model. An anime Lora won't always make the images anime-style but more often will change the gens in interesting ways (such as mood).

More information on the model and usage tips coming soon...

2DN-Pony v1.0 ReAL

Update: ReAL is now in the site generator

Update: the ReAL version ups the realism, especially with the “realistic” tag. Now produces a wide variety of images. It is being released under early access. Buzz will go towards training on a large realistic dataset. Big thank you to everyone who is dropping buzz on this model ❤️

The ReAL version is seed-matched with v1 - meaning you can try regenerating your v1 images with the same settings + seed and the general appearance should be similar in terms of content and composition.

Motivated by not being able to do certain stuff in Flux. I actually prefer promoting with booru tags as well, instead of long natural language prompts. I think the Pony base model is still unmatched.

Original description:

See sample images for prompt ideas. The model can produce anime to 80% realistic content.

If you are after a fully realistic Pony model I recommend Godiva. For a 2D anime model, check out CashMoney.

Update: Works with onsite image generator!

Thank you to Pony Diffusion team for the game-changing Pony base model.