Use this Embeddings in the Prompt section.
There is another Part of the this Embedding that can be downloaded from here. LINK
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.
๐ Like this model to keep me motivated and inspired to create more!
๐ฌ Drop a comment and let me know what you'd love to see next.
๐ Review this model to help me improve and make even better creations.
๐ Hit that notification bell to stay updated with my latest models and updates!
For Best Results use with my Detail sliders and my Checkpoints
Description
Positive Embedding for Illustrious XL.
Use it in Your Prompt Section, Keep it at the beginning of the Prompt.
Enhances your Image with Realism, details and better anatomy, Less Noise.
Use this Embeddings in the Prompt section.
There is another Part of the this Embedding that can be downloaded from here. LINK
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.
FAQ
Comments (8)
What software were you using on these ladies?
A1111
These two your embeddings broke my SD forge twice. :((
Really ! ๐
this is the first time i am reading this. Kindly explain , what happened, the errors, and more.
Stable_Yogiย I don't know exactly what your files do, but they ruin the quality of the generation (a lot). I noticed when comparing the metadata of the image generated before inserting and using your files with the same image generated afterwards that the image generated after inserting your files (in addition to being worse) had a difference in the metadata. The difference was that the metadata of the next generation indicated "Token merging ratio" values of 0.7. I had never changed those values in my SD settings, and it was strange that they appeared in the metadata parameters of the second image. This metadata of the next image should have been the same as the image before inserting your files... but that wasn't the case... your files changed something. I went into the SD settings to change/reset the "Token merging ratio" value from 0.7 to 0 and tried regenerating the same image, but what happened was that the image quality didn't improve, but there were no references to "Token merging ratio" in the image's metadata. I tried changing other settings and running other tests, but nothing helped... each generation was worse than the ones generated before I inserted and used your files. I'm sure the problem is caused by your files because after my first SD became unusable, I deleted it and created a second SD clean folder... I tested that too, and it worked fine until I inserted and used your files...
Noire90ย Thanks for sharing your experience ๐
I totally understand how frustrating it can be when something feels off โ especially when experimenting with new files.
That said, just to clarify: embeddings or LoRAs cannot change core settings like โToken Merging Ratio.โ They donโt touch global configs or manipulate your SD install.
Most likely, what happened was that an imported image or workflow temporarily included TOME values, and they carried over into your next generation session. This can happen if SD doesn't fully reset between sessions or you load a JSON with advanced settings baked in.
Also โ deleting your SD install and reinstalling from scratch is rarely needed. That step alone suggests some SD basics may need more exploration before drawing conclusions.
Iโm always happy to help troubleshoot if youโre stuck โ just tag me or reach out. We all start somewhere, and the goal is to grow together. ๐
Excellent Embeddings. I use FORGE and have utilized both positive and negative Embeddings with models such as:
novaFurryXL,
illustrij,
waiNSFWIllustrious_v15,
waiNSFWIllustrious_v14.
In all the models, the Embeddings produced outstanding results. Across all the models, the images come out similar to the semi-realistic tag, but much higher quality and more visually pleasing. The results vary slightly between models, but overall, they are wonderful. To achieve a similar level of quality, I usually need to include more than 20 tags in each Prompt, which adds extra strain on the model. Moreover, not all tags are always interpreted correctly by the model.
Overall, thank you for your work
Thank you for the feedback and testing. Much appreciated.
Details
Files
Stable_Yogis_Illustrious_Positives.safetensors
Mirrors
Stable_Yogis_Illustrious_Positives.safetensors
Stable_Yogis_Illustrious_Positives.safetensors
Stable_Yogis_Illustrious_Positives.safetensors
Stable_Yogis_Illustrious_Positives.safetensors
323_Stable_Yogis_Illustrious_Positives.safetensors
Stable_Yogis_Illustrious_Positives.safetensors
Stable_Yogis_Illustrious_Positives.safetensors
Stable_Yogis_Illustrious_Positives.safetensors



















