What? Isn't this the right time for this, given that Ash's farewell as the protagonist of the anime was just a short time ago? Oh well... you'll overcome this someday.
ANYWAY! Despite being released soon after, this model isn't related to Kirisuna (believe me), this one is an older idea. A little context for those who are confused (Links below may have "some" NSFW content, proceed with caution), this is also credits for the idea.
▼▼▼CONTEXT STARTS HERE▼▼▼
This is obviously the fusion of Ash Ketchum and Dawn. Why? Good question... But this is more common than you think. Some people believe that Dawn has the best synergy with Ash compared to all other companions because they have similar personalities, as if Dawn were Ash's coordinator counterpart, and this is more or less recognized in Pokémon Journeys with Alt. Ash and Alt. Dawn switching roles. Perhaps that's the reason. I honestly never saw them that way, as Dawn always struck me as too "feminine" and "vain with looks" to be compared to Ash's "sloppy" personality. But if it serves as an excuse to merge the two, who am I to say otherwise?
The oldest example on the internet of this idea is from 2020 with TSFSingularity's web comic. The story is about Mewtwo suddenly using its powers to merge Ash and Dawn. In this version, Ash is in control of the body while Dawn's mind can do nothing but experience Ash's actions. It's an interesting comic at first but becomes lame in the second half as it turns into a standard Ash and Dawn hentai, with the fusion losing its story focus.
However, according to TSFSingularity, the idea is even older and comes from an artwork by Rhaokja. I never found this art, so I don't know what it looks like or its date.
Some time later, hide1001 came up with the same idea. hide1001's gallery is focused on fusions, not always the most unorthodox ones. The majority are pretty normal with female-female fusions, but some examples between Ash and Dawn appeared here and there. It had its first main role in the 2021 "comic" with the rough translated title "Beware of Hex Maniac when exploring the ruins". It’s a story using screenshots from the game Koikatsu! and it's about Ash and Dawn getting lost in the ruins of a city. They end up finding a crazy Hex Maniac who casts a spell on them, consequently merging their bodies and creating a new entity called "Satori," hence the name (Satoshi + Hikari. I have no idea how "Ash" and "Dawn" would be). Unlike TSFSingularity's version, Satori is apparently a combination of their personalities rather than one of them being in control.
The idea is interesting, but unfortunately, all the dialogues are in Japanese. A version for a western audience was posted on hide1001's deviantart, but without the dialogues, which doesn't help much. Google Translator did a mediocre job for me to understand the context: It sounds like a story of Ash promising to protect Dawn but ends up failing when they find the Hex Maniac, leading to their bodies merging into this new person... A bit tragic, isn't it? Or a bit Jojolion.
I don't know if hide1001 was directly inspired by the TSFSingularity version, but it sounds like it because both versions of "Satori" are very similar physically, with the difference that the TSFSingularity version has Dawn's blue eyes while hide1001's version has Ash's brown eyes. The latter is a bit more equal in traits distribution between Ash and Dawn.
Overall, it's pretty much a slightly tanned Dawn with lines on her cheeks, spiky locks on the sides of the hair that are reminiscent of Ash's hair, and is a herm (The big breasts apparently came as a fusion "boost" in both cases). Both versions are technically that physically.
▲▲▲CONTEXT ENDS HERE▲▲▲
Now to what really matters.
[V2]
This model is a little better. It's still a bit difficult to keep the correct clothes and can generate random different ones, especially Dawn's dress. Only the cap maintains itself consistently regardless. However, the quality of the images is much better now, being able to generate several scenarios instead of insisting on the same ruined city.
[V1]
I make it clear that this model is pretty bad, the worst I've done so far. There weren't enough resources for the character, even when trying to complement it with some Dawn and Ash artworks to make the clothes consistent. The result was bad no matter how many repeats and epochs I put in.
The only reason I decided to post this model is because it becomes much more acceptable if you use large steps to generate the image. A few steps, like 20~30, will inevitably always generate an image with a messy face. On the other hand, you don't need to lower the weight much; 0.9 already generates images in good quality, but you might have to lower it a little more if you want better variation in scenarios (putting "hxruins" in the negative prompt also helps).
"Ash's cap" leaves Satori's hair the way it should be, but without the hat, the image can sometimes misinterpret the spiky hair as some sort of "French braid." "Ash's cheek marks" technically should make the cheek marks show up, but instead it just makes the face look messy. The results are better without it.
Just use the V2, really. If you can, show some love for our little... girl...? Boy...? Our little trainer and coordinator from the Pokémon world!
The King's journey is endless, seeking the best Queens.
Description
FAQ
Comments (3)
This is nice! Ash's red hat goes well with Dawn's blue hair.
As someone that makes character fusion loras I really appreciate this project. I'm looking forward to your future Fusion projects.
What images are in your Dataset? For this one did you use images of Ash and also images of Dawn separately or did you only use images where they were already fused?
I usually fuse my characters by using two loras in the same prompt to make the dataset but I'm curious about new techniques.
I used the two projects mentioned in the text, which was easier to make a Lora character, two or three images of Dawn and Ash separated for Lora to try to better understand the concept of the clothes. But Kirisuna for example, was necessary to use images of Kirito and Asuna separately with Lora mixing the characteristics when you describe a part of one and the other.
This is more about making use of a limitation of Stable Diffusion. Although you can make a model containing two people, there's a high probability of generating an image with characteristics of both. Of course, it's also possible to do fusions using two separate loras, but you would probably have less control of what would be mixed.
@TheKingMultiversal Ok cool that makes sense. It aligns with what I've learned from my fusion projects.
Definitely agree that you can't just use two loras that by itself is extremely unstable lol
However its helpful if you need more images in your training data you can use loras to get more images for training. I used this technique to make my OCs: Stardust and Toni, and then I photoshopped the generated images to make the clothes perfect.
I'm looking forward to more of your Original Characters/Fusion projects.
I think there is a lot of great potential.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.






