Banana Splitz XXL - V1.1
(Banana XXL + FurrySpec | Simple Furry Mix + ReKemonoV9 + Furry Dreams + a few LoRAs )
Minor change in prompt adherence
style update - slightly more painterly and lite 2.5D aesthetic.
Prompt adherence slightly LESS sensitive (allows for more advanced prompting)
Camera angles and poses are about the same.
Banana Splitz XXL - V1.0
(Banana V9 + FurrySpec | Simple Furry Mix + DeepNavyV3+ ReKemonoV7)
Minor update to coloring + lighting
Major style update - smoother, cleaner, less painterly more solid.
Prompt adherence slightly more sensitive.
Offers a high quality 2D - 2.5D look with expressive posing and faces.
Camera angles are more creative when propmting for dynamic angles.
Banana Splitz - V9.0
(Banana V8 + V6)+DeepNavyV3+Luno V6[Renamed ReKemono])
Major update to coloring, lighting, poses, camera angles.
LoRa compatibility is still great
Minor clip difference from 8.0 (adherence is roughly the same)
Banana Splitz - V8.0
(Banana V6+V7a+V3)+DeepNavyV3+Hyphoria+More)
Major clip update. Expanded clip with Hyphoria
which is a model that cleverly merged 10 different models clips
Rule 34 Data set added as usual.
Prompt adherence is crazy strong.
LoRA compatability is also very high.
Backgrounds are wildly beautiful.
Can do humans/Anime with no problem.
Extremely large spectrum of styles can be prompted.
You can combine artist tags easier like in the SD1.4 days
_______________________________________________________________________
Banana Splitz - V7.0 Side B
This is basically V7 but with a more westernized toon style
(Banana V6+V7+V5(ReMerge)+DeepNavyV3+Oops All Toons)
LoRA Used:
KSM - Gives the model a more bit of western toon appearance. Helps with artist tags
Dirtysoda - Another LoRA for helping westernize this model more
Rule 34 Dataset LoRA - Smallest ammount used for activation. This is for better LoRA compatibility and higher usage causes noise. To finetune the R34 Lora within this model, please use actual Rule34 Lora. (READ INSTURCTIONS)
https://civarchive.com/models/1191424/pornmaster-rule-34-prompts-illustrious-and-noob
__________________________________________________________________________________________________
Banana Splitz - V7.0 Side A
(Banana Mix+MolKun-Mashup+Luno Mix)
Banana Splitz XL 6.0 + Banana Splitz 5.0
Other models in recipe
Hyphoria
Nlxl:
Nlxl v0.9 is a checkpoint that produces more stable and detailed results through enhanced Model Merged checkpoints created through model merging, combining the natural image generation capabilities of Illustrious, the extensive dataset of Noob, and the unique strengths of different models depending on the version
_______________________________________________________________________
Merged LoRA info, so you know how much extra of the same lora to use, or if it lora's wont be compatable. Also lets you know whathey do.
KSM - Gives the model a more bit of western toon appearance. Helps with artist tags
Rule 34 Dataset (LoRA)
The developer speaks Chinese only and doesnt really reply correctly to anyone.
but this LoRA adds a large amount of tags that do not exist on danbooru or e621
This creates an effect that allow an awesome solid mix of tags/natural langauge.
It also makes NSFW content so much better. I think it also re-enforces the knowledge of already know tags for pose, expression and anatomy wise.
bruhhhh Its such a huge an potent lora. Its 3.4gb and is merged is sometimes merged in SUB 0.1 weights. like 0.08, 0.06. Its gonna be doing something powerful
__________________________________________________________________________________________________
Banana Splitz - V6.0
(Banana+Oat+Sudachi Mix)
-Banana Splitz XL 2.0 & 4.0 mix
-Sudachi XL
-Plant Milk - OAT
-Xavier-Shift (Trunk of Furry Junk)
-Mol_Kuen DeepNavy
_______________________________________________________________________
Banana Splitz - V6.0 BETA
(Banana-Hassaku+Walnut Mix)
***NOTE V3.0 AND UP IS FOR EULER A SAMPLER ONLY**
New base merge recipe:
-Banana Splitz XL 3.0 & 5.0 mix
-Sudachi XL
-Plant Milk - Walnut
-Hassaku 3.0
-Mol_Kuen DeepNavy
-Hyphoria
-Rule34 Prompt Lora updated to V2.0
__________________________________________________________________________________________________
Banana Splitz - V5.0 [ READ ME! ]
(Banana-Hemp+Walnut Mix)
***NOTE V3.0 AND UP IS FOR EULER A SAMPLER ONLY***
Try using a CFG Rescale plugin for ComfyUI or A1111 if want to attempt another sampler
New base merge recipe:
-Banana Splitz XL 2.0 & 4.0 mix
-Sudachi XL
-Plant Milk - Hemp ll
-Mol_Kuen DeepNavy
-Rule34 Prompt Lora updated to V2.0
TEXT ENCODER MODIFICATION:
This version will properly acknowledge prompting from those who write longer and wildly detailed prompts. BUT! it will also help/support simple prompts into creative works of art!
Great for beginners and fantastic for pros to push their creative limits!
Positives for the Pros!
highly specific prompting will now have a much higher chance to affect small details.
I suggest even putting weather or time of day (morning, golden hour, night).
Even objects in background like a swing on a tree or a bookshelf indoors.
Style/Artist blending can have surgical precision.
detailed negative prompting can help shape result moderately better too
Positives for everyone!
Simple prompts will have a boost too:
Short or vague prompting will give creative wildcard like behavior.
Can give good solid and creative prompts with simplistic prompting.
Good for beginners to help build upon their prompts making them more verbose and gaining the skill to write more detailed prompts.
Great for beginners to see the changes by adding, changing how they prompt.
Remember! Everyone prompts differently and this model can certainly help you tone up on your own prompting skills!
_______________________________________________________________________
Banana Splitz - V4.0
(Banana-Walnut Version)
New base merge recipe:
-Banana Splitz XL 2.0
-Plant Milk - Walnut
-Sudachi XL
-Rule 34 Dataset V1.0
-Style is mostly the same, but higher quality.
-More detailed backgrounds
-Prompt adherence is average. Needs work.
-Extremely LoRA compatible
-Artist styles absolutely SHINE on here.
-Can one-shot generate a good image 3/4th of the time
_______________________________________________________________________
Banana Splitz - V3.0
(Banana Flax Daquiri Mix)
A 'high-quality' Western-Kemono hybrid with crisp linework, soft but dimensional shading, and a balanced 2.5D aesthetic.
Better and easier prompt adherence
Clean, stylish, thick lines with glossy 2.5D textures
Contains Rule 34 Dataset.
Great eyes! Great Fingers, Awesome poses.
Check out this amazing finetune of Banana Splitz by LS582.
Fifth Mix - Banana Daquiri does exactly what its told, clean, crisp style. This is what ChromaXL should have been IMO.
If you like this model try it out!
_______________________________________________________________________
Banana Splitz - v2.0
(Banana Daquiri Mix)Updates:
-Moderate Style Change
Model now is has moderately higher saturation, slightly lower contrast, more colorful.
Less flat and washed out (if you like that look from V1 I may consider making a V1.X, Plz Comment)
When no style is specified you will get random styles beetween 2.5D cartoon and 2.5D semi-photorealstic
-Massively Improved prompt adherence.
Has improved natural language understanding(needs to be mixed still with tags)
Prompt this model like a combo of Fifth Mix - Banana Daiquiri and NoobAI
-Refined quality and body comp.
Much less random zoom ins and cropping when using "close up" tags
You can get good eyes on raw gens 60% of the time with full body/multi character
You can get good eyes on raw gens 95% of the time with portrait/close up/solo
Different male and female body types
better knowledge of animal dicks and pussies(specifically canine)
-Pure NoobAI cyberfix, train difference injection
Use e621 tags for artist associated with NoobAI
vast knowledge of furry artists and can nail the style very closely 80% of the time. (check last 3 images in example images)
Natural Langauge is better understood. but not perfected.
-Making Lora/Lyrcoris models use the following.
NoobAI EPS1.1
Illustrious V0.1
WAI-NSFW-Illustrious V11 (Smooth looking with mild risk of sucking)
Fifth Mix Or Banana Splitz (Highly creative / Lower risk of sucking, but makes it useable for these models only)
__________________________________________________________________________________________________
Banana Splitz - v1.0
(The Original Banana Splitz Mix)
ORIGINAL VERSION 1 MERGE DETAILS:
Models used:
Fifth Mix
FurryToonMix
MolKun Mix
NoobAI CyberFix EPS V1.1
Custom Oops! All Furries
(Re-Merge/Pruned All trace of models that use ChromaXL Or ChromaMorgy
Custom Indigo NoobAI Mix(I might release this)
Custom Nova Furry Mix
(Custom Add Diff/Train Diff Mix of 3 versions - Might release this too)
Hassaku V2.1
Interested In any of the custom models used?
Let me know!
Description
FAQ
Comments (33)
Great Model m8
NOTE: IF YOU SEE YOUR PROMPT USED IN COVER IMAGES DM ME FOR COMPENSATION!
Just tried a few and I gotta say you absolutely cooked with V5, the prompt adherence is amazing and the overall clarity seems super nice
The v5 it really takes care of the enviromental details, I'm liking this one a lot! ♥
Thankies for this update ~
Amazing. As always, this is simply better. Background detail is better, so is character heeding. Especially things like character poses and fur/eye colors.
but I do have a bit of an issue. (Probably just personal) most of the, the eyes I get are blurry, or just look off. I use all the quality enhancing tags that you use in your prompt. But yeah, probably my issue. (Edit; turns out fat prompts, which are basically what I always use for gens, make the image more blurry, ruining details.)
Overall, best version so far. 10 Muskets Out Of 10 Black Forest Cakes.
Thank you very much for the detailed review.
Long prompts have the tendency to throw off the eyes and fingers a bit.
are harder to do and it takes a quite sometime to get in tune with the text encoder settings for each model.
Ill give you a good guide to make you into a prompting pro for my model merges.
First lets understand what a checkpoint trained model is vs a checkpoint merge
Checkpoint Trained models mean someone took illustrious and trained 100k to 500k images more into it. Just like Illustrious was made by training millions of images on SDXL.
1. All my models contain these TRAINED checkpoints illustrious 2.0 + NoobAI 1.1 EPS.
2. All my models contain a huge 3gb lora that adds Rule 34 dataset.
Now illustrious 2.0 has some pretty decent natural language prompting, but should be padded with tag style prompts
Also more verbose tags from rule 34 blend amazingly well with e6/danbooru tags creating an illusion of better natural language. Ontop of that, any tags thats overlap between r34/e6/booru are re-enforced to be even better.
Now in the next comment im going to use one of your prompts and re-work it for you so that its optimized to be best.
This is the image I chose
https://civitai.com/images/90305739
Positive Prompt
Lets start by breaking them down into sections.
First start with quality tags
masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, 4k, absurdres, newest,
So you had a few tags that repeated twice, also the more quality tags you add, the more it pushes from 2D to 2.5D to Semi-Realism.
Also doing this... (furry:1) is the same as just writing furry. So you dont need to use () unless youre gonna be doing 1.1 or above, because each tag/token is weighted already at 1.0 by default
Next setup characters and details
This is how I would modify it. but you mostly did a solid job on this, you dont really need to include nude twice, but it should be fine to do that.
Anthro, Furry 2boy, male/male, fox, Orange Fur, blue eyes, pecs, nipples, Nude,
weasel, brown fur, Nude, Penis, Balls, Oral, Oral Penetration, Tongue Out, (Thick Tongue, Long Tongue) Tongue Wrapped Around Penis, Carressing Balls, Beach, Coastline, Waterfront, Sunny
Sometimes if you want to describe to characters you can group them togeher using the parenthesis.
Example: Two characters with different clothing, eye colors, species
(otter, brown fur, blue eyes, open flannel shirt)
(fox, orange fur, green eyes, black undershirt)
this also works when wanting to give one an action like performing fellatio.
ex:
(Fox, orange fur, green eyes, nude, oral penetration, fellatio, tongue out)
(otter, brown fur, blue eyes, nude, penis, balls, sitting on sofa, moan, pleasured)
________________________________________________________________________________________
For your negatives though you making them TOO detailed. Illustrious only needs a few simple negatives.
1. Dont use pony scores!
2. Dont put weights of 1.8 on these worst quality, bad quality, text, logo, blurry, bad anatomy
its epescially risky to do that with embeddings like bad-hands-5 which I dont even think is for illustrious.
I highly suggest you use these embeddings for positive and negative
https://civitai.com/models/1302719?modelVersionId=1833157
This will help you so you dont have to write out all the quality tags and negative prompt qualities.
@LatteLeopard Amazing Info, I’ve modified my prompts (more specifically negative) over the days. Would weights be needed? Is “2boy” needed or will “duo” work just fine? and would using line breaks
(like this for example)
be necessary for prompts? (I’ll post an image for reference of my prompting, furthermore, I’ll be using the example you gave)
@M1Carbinemann for universal use, if you want to try describing 2 male characters with one tag. I would try "2boy" because it assumes 2 males, and a lot of models have heavy female bias. So you might end up with a female/male pairing with just duo.
as for BREAK you can use that in A1111(or comfyui) to properly separate two chunks of tokens.
Once you go over 50 tokens, it begins a new chunk of data so like if you were describing a character late in the prompt and half of the details when over 50 tokens, it would split the excess into the a separate chunk.
This increases chances of the AI losing its train of thought and doing weird stuff.
so to prevent this you can use BREAK, to end the chunk early so you can properly prepare for it using the token counter you see in the prompt windows.
Then you can end the chunk of tokens at 40. so when you start describing the character it will all be in that second chunk of tokens and not jumbled between two.
@LatteLeopard A1111? How do I explain I use this site because it’s the ONLY thing I have. (Best tech I have is a Chromebook) regardless. Thank you for all this info. I’ll be sure to use it to it’s fullest.
@M1Carbinemann if you are using civitai's generator dont worry about using BREAK. it will automatically do what break does.
@LatteLeopard Phew. (I suppose I don’t use it) You’re the Goat! Thanks again.
"glowing eyes" in negative tend to make non blurry eyes 😁👌
@SXCD2719 Thanks!
perfect!
Thanks dude!
I typically use Euler sampler, which worked fine on v4, but on v5 i get distorted mess with Euler. I changed to Euler ancestral and it was fine. Is that expected?
Yes! That is absosultely expected with Euler NON ancestral and dpm++ 2s Karras.
I am impressed you were ever able to work it with V4!
This is because the newer ecosystem of illustrious(NoobAI) models seem to merging in LoRA's that are meant only for Euler A. Then as more models merge with those, people are unkowningly spreading this optimization for Euler A.
I didnt even realize this was an issue until my release of IffyMix V4.
Also if you are a ComfyUI user you can use
dpmpp_2s_ancestral_cfg_pp if you use the CFG Rescale
the CFG Rescale should also work for the Euler/DPM++ 2S samplers
@LatteLeopard Ty for the tip! Your model is great for sure. I find it to be the best IL model for the clean lines and classic "toon" furry style. Keep up the great work!
@Foxbite Thank you so much!
I appreciate the feedback. Hopefully this helps other people not using Euler A as well.
Maybe for V6 I will try to make it compatible with Euler(non ancenstral) and DPM++ 2s Karras
@LatteLeopard Euler non-a was my default for V3 and V4, but it randomly started tweaking when I tried V5 and that left me a bit confused for a bit haha
Hey @LatteLeopard nice model,
I've heard really good things about this model in terms of flexibility and lora compatibility and style adherence. Quick question, do you have any recommended negatives, and settings for your model?
This is quite basic but works amazingly for me, i've been using this for a lot of time
"worst quality, old, early, low quality, lowres, signature, username, logo, bad hands, mutated hands, glowing eyes, simple background"
You can even improve by adding this for a better 2.5D, 3D or realistics outputs, it's optional ofc
"(thickline work, 2d, flat coloring, flat shadows):1.4,"
Good luck and have fun there! ❤
Exactly what SXCD2719 said.
If you follow her guide, you will have a great gens not only for my merges, but for all illustrious merges!
Question: Should I be all capping my prompts? (I just need help with formatting)
I feel like doing this just messes something up, I don’t know what.
(For Example, what I put at the start of any Image I generate) “Kemono, Digital Illustration, (Masterpiece, Best Quality:1.4), Newest, (Detailed Background:1.1) (Extremely Detailed, Intricate Details) Detailed Fur, Messy Fur, Shaggy Fur, Solo, Anthro”
You can try with the very basics
"masterpiece, best quality, amazing quality" with "anthro" or "furry" if needed
Then you can start adding one per one tags, even try new ones for experimenting
Sometimes i use this for realistics aproach
"masterpiece, best quality, amazing quality, very aesthetic, ultra-detailed, 4k, highres, volumetric lighting, ray tracing, realistic, furry, anthro,"
Sometimes i use for highres ones (mostly for local comfyui aproaching to a more scenery setting)
"masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, absurdres, newest, scenery, furry, anthro,"
Sometimes i just use the basic one
It always depends on what you're looking for!
Best wishes! ❤
1.4 is quite a strong wieght, and when you apply it to the quality tags. it sometmes "overbakes" them giving it a paradoxial worse result.
its best to not wieght any of the quality tags. just use more tags that describe quality
SXCD2719, explains it like a pro.
Follow her advice, and you will see a deffinerence.
Also remember to use Euler A or LCM only!
Also if you dont have access to inpainting, and you focuse heavily only background details. IT will focus more on making the scenery pretty than the faces and eyes
Banana Splitz has become my main go-to furry checkpoint aside from the perpetually cute Kiwi Mix. ❤️
Sorry, Nova Furry, I found someone new. 😮
Wow I really appreciate that, but Nova furry is still good to use!
Remember our models are about different styles, you should treat all our model merges like different creating art with different tools.
Check out reed_furry_mix and Xavier.
They offer really unique styles too!
@LatteLeopard Oh, I know. I do dabble with other checkpoints, but I tend to always circle back to the ones I already really like. I'm a creature of habit and all that. ;)
@anomedra im so glad to hear that. I appreciate your support very much! thank you!