PhotoPedia XL 📷

Update - 19 / November / 23

Further increased realism, a lot more small details. Model works good on wide range of concepts.

I will be updating this info section soon.

📌 Merge info / recipe

1️⃣ Type 1 ✔️First merge

• CinemaXL + AltXL + XL6 = M1 ( 3 Sum / Normal / MBW )

• YamerXL + Fuduki + RealityCheck = M2 ( 3 Sum / Normal / MBW )

• CrystalClearXL + CherryPickerXL + MiaModelXL = M3 ( 3 Sum / Normal / MBW )

• M1 + M2 + M3 = M4 ( 3 Sum / Normal / MBW )

• Yamer + M4 = PhotoPedia XL Type1 ( Weight Sum / Train Diff / MBW ) ✔️

2️⃣ Type 2 ✔️Fine blend between digital / realistic

• RealCartonXL + RealVisionXL = M21 ( Sum Twice / Normal / MBW )

• LEOHello + XXMixXL = M22 ( Sum Twice / Normal / MBW )

• PhotoPedia XL + M21 + M22 = M23 ( Sum Twice / Normal / MBW )

• M23 + M21 + M22 = PhotoPedia XL Type 2 ( 3 Sum / Normal / MBW ) ✔️

3️⃣ Type 3 ✔️ Increased realism / Slightly more NSFW

• PhotoPedia 1 + XXMixXL + PhotoPedia 2 = PP1 ( 3 Sum / Normal / 0.45A - 0.27B )

• PhotoPedia 2 + PP1 + RealVisionXL = PP2 ( Sum Twice / Normal / 0.57A - 0.32B )

• PhotoPedia 2 + LEOHello + RealVisionXL = PP3 ( Sum Twice / Normal / MBW )

• PhotoPedia 2 + Devil + JuggXL5 = PhotoPedia XL Type 3 ( Add diff / Normal / 1.0 A )

4️⃣Step 4 - Manual MBW WIP / Unreleased

• PhotoPedia 3 + JuggXL5 = PE1 ( Weight Sum / cosineA / MBW )

  • alpha,beta : ( 0.8, 0.25 )

    • weights_alpha : [ 0.7, 0.8, 0.8, 0.4, 0.2, 0.9, 0.2, 0.4, 0.5, 0.5, 0.9, 0.7, 0.5, 0.6, 0.3, 0.9, 0.2, 0.4, 0.9, 0.8, 0.2, 0.3, 0.4, 0.6, 0.7 ]

• PE1 + RealVisionXL + PE1 = PE2 ( Sum Twice / cosineA / MBW )

  • alpha,beta : ( 0.8, 0.25 )
    α，β：（0.8，0.25）

    • weights_alpha : [ 0.7, 0.8, 0.8, 0.4, 0.2, 0.9, 0.2, 0.4, 0.5, 0.5, 0.9, 0.7, 0.5, 0.6, 0.3, 0.9, 0.2, 0.4, 0.9, 0.8, 0.2, 0.3, 0.4, 0.6, 0.7 ]

    • weights_beta : [ 0.5, 0.4, 0.6, 0.6, 0.4, 0.4, 0.6, 0.5, 0.4, 0.5, 0.5, 0.6, 0.4, 0.6, 0.6, 0.3, 0.7, 0.6, 0.4, 0.6, 0.4, 0.6, 0.4, 0.4, 0.6, 0.7 ]

• PE2 + JuggXL5 + PE2 = PE3 ( Sum Twice / cosineA / MBW )

  • alpha,beta : ( 0.8, 0.25 )

    • weights_alpha : [ 0.7, 0.8, 0.8, 0.4, 0.2, 0.9, 0.2, 0.4, 0.5, 0.5, 0.9, 0.7, 0.5, 0.6, 0.3, 0.9, 0.2, 0.4, 0.9, 0.8, 0.2, 0.3, 0.4, 0.6, 0.7 ]

    • weights_beta : [ 0.5, 0.4, 0.6, 0.6, 0.4, 0.4, 0.6, 0.5, 0.4, 0.5, 0.5, 0.6, 0.4, 0.6, 0.6, 0.3, 0.7, 0.6, 0.4, 0.6, 0.4, 0.6, 0.4, 0.4, 0.6, 0.7 ]

• PE3 + RealVisionXL = PhotoPedia XL Type 4 ( Weight Sum / cosineA / MBW )

  • MBW - RING08_5

    • weights : [ 0.2, 0 ,0, 0, 0.3, 0.3, 0.4, 1, 1, 1, 0.3, 0.4, 0.4, 1, 0.8, 1, 1, 0.4, 0.2, 0.2 ]

   

• Initially, I wanted to test AutoMBW for this step, but I decided to leave it for another time, it takes way too long. Instead, I went with manual block editing.

• If the weights look silly, they probably are x) As long as it improves the overall quality!

5️⃣Step 5 WIP / Unreleased

• PhotoPedia XL Type 4 + LCM

6️⃣Step 6 Unreleased / No ETA

❗ Further down the line - possible fine-tuning on own dataset

💭 Although the end goal of the merges is to reach a certain level of realism, I do recommend you to try all the versions provided, they have slight differences, but I guess certain versions might be more appealing to your eye. I personally cannot decide on which is my favorite so far, they all have their uniqueness.

📣 GENERATION PARAMETERS & SUGGESTIONS

ℹ️ I ran most of the generations with the following settings

Sampler - I used many. Try running DPM++ 2S a Karras Steps - 65 - Can go lower / higher, depending on your sampler. CFG - 9 - You probably might want to go lower on this. Hires upscale: 1.5 Hires steps: 15-20 Hires denoising: ~ 0.5 Upscaler - ESRGAN 4x / 4xUltraSharp I have not used ADetailer so far with this merge, having some issues with it. I do recommend it though. I didn't inpaint generated images. I didnt use any negative embeddings. I suggest testing on your own to see which one you like the most as some might change the image more than others. I used Detail Tweaker LoRA usually around 0.3. Upscale in A1111 in Extras OR Upscale outside A1111 - Using Waifu2x GUI

ℹ️ Prompt Example

amateur photo of a --subject-- --doing something-- / --wearing something--, analogue, raw, polaroid, looking at camera, (from side:1.2), detailed skin texture, natural skin texture, perfect eyes, perfect iris, high detail eyes, detailed iris, detailed cloth texture, detailed facial features, aesthetic, depth of field, cinematic light,(high quality, best quality:1.3), Extremely high-resolution, Kodak Portra 400 camera f1.6 lens, Fujifilm Superia Premium 400, Nikon D850, <lora:add_detail:0.33>

📌 More info will be added later