Important Note :
Use Positive embeddings in the Prompt section
Use Negative embeddings in the Negative Prompt section.
For guidance, check the images in gallery below.
Download New Versions here :
Stable_Yogis_PDXL_Negatives2 :> https://civarchive.com/models/1331758
Stable_Yogis_PDXL_Positives2 :> https://civarchive.com/models/1331980
Join me on my Patreon for exclusive perks and early access to unique resources.
To discuss custom LoRa's or models, feel free to connect on Discord.
👍 Like this model to keep me motivated and inspired to create more!
💬 Drop a comment and let me know what you'd love to see next.
🌟 Review this model to help me improve and make even better creations.
🔔 Hit that notification bell to stay updated with my latest models and updates!
1: NEG_EMBED_STABLE_YOGI V1 (For SD1.5, Too old now - please switch to V3)
2: NEG_EMBED_STABLE_YOGI V2 (For SD1.5, Enhanced Negative Embedding)
3: Low_Quality2xx_ By_Stable_Yogi (For SD1.5, Quality Enhancer Negative Embedding)
4: Low_Realism_2xx_By_Stable_Yogi (For SD1.5, Realism Enhancer Negative Embedding)
5: Realism_Quality_2xx_By_Stable_Yogi (For SD1.5, Realism + Quality Enhancer Combined Negative Embedding, Low_Quality2xx_ By_Stable_Yogi + Low_Realism_2xx_By_Stable_Yogi)
6: Morenegative_Stable_Yogi (For SD1.5, Bonus Negative Embedding!, This was for personal use)
7: NEG_EMBED_STABLE_YOGI V3 (For SD1.5, New Updated Negative Embedding - can be used with Morenegative_Stable_Yogi, Low_Quality2xx_ By_Stable_Yogi, Realism_Quality_2xx_By_Stable_Yogi, Low_Realism_2xx_By_Stable_Yogi for regular images, please test the image quality and decide which embedding are giving good results together. For only female photography use this embedding with influencer tools , see image examples.)
8: Stable_Yogis_PDXL_Negatives (PDXL Negative Embedding + only for realistic Image, Dont use With Anime, Cartoon or Art Styles)
9: Stable_Yogis_PDXL_Positives (PDXL Positive Embedding, keep this at beginning of your prompt, + only for realistic Image, Dont use With Anime, Cartoon or Art Styles)
10: Stable_Yogis_Animetoon_Negatives ( Good with Anime, Toons, and realistic images )
11: Stable_Yogis_Scenery_Positives ( Great for realistic scenery images )
12: Stable_Yogis_Scenery_Negatives ( Great for realistic scenery images )
For Best Results use with my Detail sliders and my Checkpoints
Description
Latest Negative Embedding for SD1.5
FAQ
Comments (34)
there's no word to describe this... AMAZING!!!
Thanks for the feedback.
What are some neg prompts to lessen the chance of low quality and to prevent 'moaning' face? Every time I try to generate a person they have a 'moaning' look to them.
((moaning:1.2)) try this.
Add the negative embedding for the checkpoint.
All of those are available on this page.
Sorry for the newb question, just want to make sure. In order to use the positive embedding, I add [Stable_Yogis_PDXL_Positives] at the start of the positive prompt? including the bracelets
Add just this = Stable_Yogis_PDXL_Positives
make sure to keep the embedding file in the embeddings folder.
Are the embeddings baked into your models like Realism or Babes? or do I still need to download the embeddings and use them separately?
Embedding are separate models.
Pls download and add them your embedding folder.
@Stable_Yogi what do you do with them after that exactly?
@draggedfromthelight859 Just leave it your prompt it handles anatomy fixes, enhances realism and more.
Hi Yogi,
New fan here. You definitely have a lots of stuff!! confusing at first but I'm getting the hang of it.
I've noticed once consistent issue though, using the V4 and V5 real_xl checkpoints with the positive2 and negative2, its extremely same-y, same faces, same lighting (dark around and heavy professional-ish/glowy lightning on the persons), and always in a field for some reason? Prompting a location is hard and the model/embedding seems very stuborn on the lightning, appearances and the location. It feels more like a lora than a model
Any idea how i can help that?
Thanks for the feedback, Please learn from the gallery images, i have 220K+ . There are no such issues,
Now judging by what Stable_Yogi has said you must be trying to move at too quick of a pace, be commanding the AI too much and expecting small details, when the AI wants the freedom to perform what it wants, too bad the images have similarities that offend you, you are not accepting of it's creativity, minor issues when there are things unseen to be cherished within the generative AI prompts, I have some really beautiful artworks that AI made for me when I got c razy and could let it flow. Come check them out!
so the ones i should use for 1.,5 are thre negative and postive ones furthest to the left in the list at the top and v3 is outdated?
V3 also works well.
These 2 are new and they work good too.
Both have lots of difference's you can try both and use the one which works for you.
@Stable_Yogi thanks friend, your work is exceptional!
@Stable_Yogi V3 works fine but these newer versions doesn't even show up in my text inversion tab, any idea wwhy?
@chavezramiroadrian XL embeddings will only show when XL checkpoint is loaded, sames goes for SD1.5
@Stable_Yogi thanks for the response yeah I loaded the model but still nothing shows up in the UI. The v3 is displayed correctly so thats really odd
im noob, how to use it? anyone has an exemple workflow?
assuming A1111, put in embeds folder, then in UI go into Tex inversions tab
@TheCAL Correct, then refresh and also choose the correct model.
if SDXL then you will only see sdxl embedding, if you choose Sd1.5 then you will only see sd1.5 embedding.
Sometimes generating aimage after loading a checkpoint helps you findind your embedding.
hi guys, sorry, also a noob, what are wildcards and embeddings nad any way to use in fooocus?
@Stable_Yogi What is the -neg for?
@Lewd_N_Geeky -neg (simply reminds you to put that embedding in the negative prompt.)
@Stable_Yogi Thanks for the help.
I've tried, a few Neg/Pos embeds, I think this one may be the real deal particular the Negative. Everyone's MMV however
tried the negative with a bad experience. makes images even worse - ignores most parts of the prompt, generates extra limbs, etc.
hey sorry for the dumb question but is there a link for a step by step instruction on how to enable this?
Hey Yogi, thanks a lot for the share. One question, which tags are included inside of your PDXL positive and negative embeddings exactly? Because without knowing it i don't know if i need to add scores or extra tags to my prompt.
You don't need to add scores, those are covered.
every time i try to use these they only output portrait compositions
This actually has a very noticeable effect. It doesn't prevent weirdly mangled fingers, but it improves pretty much everything else.