shirotora84
New Member
- Sep 4, 2019
- 4
- 5
Oh, I have no problem telling people to go fuck off if I decide I don't want to do something anymore. I'm willing to help where I can. My offer stands. Take it or leave it.
- Thread starter Novilon
- Start date
Pleevy
Member
- Nov 23, 2024
- 232
- 730
Oh, no question, I'll take it. I never meant ot be "the writer" but, progress has been made. DM me if/when you want and we can go over stuff.
Novaca
Newbie
- Feb 19, 2022
- 83
- 292
Well, I guess I owe you some explanations for disappearing and then coming back out of the blue.
Well, just like our friend Pleevy, I had some health issues and took some time off from pretty much everything.
Now, if everything goes well and I can get back into the swing of things, I'll start taking pictures again.
Well, just like our friend Pleevy, I had some health issues and took some time off from pretty much everything.
Now, if everything goes well and I can get back into the swing of things, I'll start taking pictures again.
shirotora84
New Member
- Sep 4, 2019
- 4
- 5
I keep running into an error when I try. I can talk on Discord, if you have it, same name different profile pic
FoxyTails
Member
- Feb 9, 2020
- 321
- 788
Right, well can't speak for everyone who been working on the graphics (newbie myself) but I scanned through the posts & come up with the following info
SWP: has stated he uses Stable Diffusion (& yes this IS a standalone) with 'several models' but mainly Pony Diffusion and KIMXL. Using SD's 'PNG Info' I dragged A SWP 'Cream clothed' image & I see that he had used 'Lora: Pony Diffusion V6 - Smooth Anime ' and a LoRA 'Cream_the_rabbit'
Novaca: Stable Diffusion with nothing except NovaFurry checkpoint
MrFluffum: ForgeUI (basically Stable diffusion with a different interface afaik) with NovaFurryXL4 and a 'Age Slider LoRA' to create Cub pictures. He did post a lengthy post as a 'simple' tutorial for prompts etc
Foxy: Stable Diffusion (A1111 version yet another UI) - currently experimenting with NovaFurryXL and ponyDiffusionforAnime 'checkpoints' - occasionally trying different LoRA models but trying to keep things 'simple' but not depending on them
As mentioned above - apparently you can copy some graphics into SD to 'get the prompts' & you can copy them into the 'Txt2Img' & click Generate, however unless you have exactly the same checkpoint/lora (Don't ask me what's the differences between checkpoint & Loras) you won't get the same picture
Using the prompt from SWP 'cream dress OK eyes up'
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
As you can see ponyDiffusion seems the closest to SWP style.
As for me - as I'm still learning I 'cheat'.
I go to Civitai (BTW I'm currently using VPN in Opera to do so atm thanks for the advice)
look at the pictures to try & see one that has the 'pose' I want
click the paintbrush icon (& hope it has 'Remix')
this then opens up the side panel & hopefully there's a 'prompt' there
I look through the prompt & try & figure out what causes the 'position' I want,
Copy that in SD with a brief description of the characters I want, then click generate' with random seed until I see one I like
You don't have permission to view the spoiler content.
Log in or register now.
I looked up about how to make your own poses as to not rely on other prompts & hope for the best, & many tutorials talk about 'controlnet' & 'openpose' but for some reason while I can get them installed & I get the 'wireframe' models showing, it doesn't work - the pose isn't applied when generating.
I'm also currently looking into something called 'regional prompter' that allows you to create region & can describe each region with a specific prompt (currently SD uses the whole prompt over the whole generation) & I may look at MrFufficum 'guide' about inpaint to see how to combine multiple pictures into one
So......
As a suggestions, to make things easier (simpler?!) I suggest we decide on a single checkpoint for the style,
Try NOT to use specific character LoRAs,
Create a set of full portrait images for each character (Naked/Clothed ?) using a simple prompt,
Generate several random one until you get a picture that seems oK - then make a note of the 'Seed'
(we're probably get better images of certain character with different seeds eg Cream could be better with 12345, while Blaze could be better with 76543)
THEN using the seed & same checkpoint we 'should' get repeatability for each character
Keeping it 'simple' could make it easier for multiple fans to generate consistent graphics even if they're newbies to AIgraphics
Last edited:
thelastkingdom36
Member
- Jul 16, 2022
- 296
- 483
I decided to give the 0.7 remake a go and see how things go. Right off the bat after the first scene in the living room I had to stop and do this post. For some reason things do not add up. It appears even if you open your mouth you get charged a energy deduction for talking. You can however go to the garage and find stuff then go to the foyer and find stuff and then go to the kitchen and find stuff without using the slightest amount of energy. Go to the mall for ice cream. After you summon Cream who should all ready be by you side cost energy. Choosing the Ice Cream cost $10 and the Icream another $10 but the math doesn not add up. SO you drop the first 10 and the second shows up as 10 buy you are deducted $5. If you watch the display on the left side of screen and watch the figures change you can see what's happening. Example go to move you 500 minus 10 for the bus then 30 for the movie leaving you with 460. Ice cream -10 = 450 then Cream's ice cream is -10 = 440. But NO! in fact according to the game it says -10 for Creams Ice Cream but deducts only $5 leaving you with 445.
All I'm saying is that I know what is involved in programming. I majored in COmputer Science and took ever codeing language and wrote programs with each one. Coding in time consuming, but you must check your program as you go for errors and workable. Common mistakes take place when you try to go to fast. It is very hard to go back on large anount of code to try and find a mistake or mistakes. Right now, the math does not add up. The other side of the coin is you'll get credit for Cream touch on one scene but then not on another. Example: Living room start you hug Cream you get touch other times you don't. In the movie you put Cream in you lap and sooth her tears. No points for touch.
Thing is in my opinion, energy should be deducted only from any activity, like sex playing games swiming etc. not from simply talking. Anyway, as I said, it's my opinion. At 77, I may be old, but my eyes and brain are very good.
All I'm saying is that I know what is involved in programming. I majored in COmputer Science and took ever codeing language and wrote programs with each one. Coding in time consuming, but you must check your program as you go for errors and workable. Common mistakes take place when you try to go to fast. It is very hard to go back on large anount of code to try and find a mistake or mistakes. Right now, the math does not add up. The other side of the coin is you'll get credit for Cream touch on one scene but then not on another. Example: Living room start you hug Cream you get touch other times you don't. In the movie you put Cream in you lap and sooth her tears. No points for touch.
Thing is in my opinion, energy should be deducted only from any activity, like sex playing games swiming etc. not from simply talking. Anyway, as I said, it's my opinion. At 77, I may be old, but my eyes and brain are very good.
FoxyTails
Member
- Feb 9, 2020
- 321
- 788
Well, I took the bullet today to try & do what I suggested about common AI graphics in earlier post.. Oh dear, it's not as easy as I thought.
I first tried 'using xxx checkpoint & use a random seed until I get a graphic I like & write down the seed' - problem is, if I altered the prompt to add/remove clothing etc... it screwed up the original graphic .... failure
I have today instead used one of the new graphics (sally hi.png) into 'pnginfo' to get a general prompt & copied that into 'txt2img', then using JUST 'ponyDiffusionForAnime' & changing the name in the prompt I get some 'reasonable' graphics - however some characters will not work
I then tried again using 'novaFurryXL_illustriousV9b' & repeated the same thing again - just changing the name - some names worked & some didn't, some names came up with naked version, but even when I added clothes to the text the graphics didn't get too badly corrupted
Anyway, the 'prompt' I used for both checkpoints is:-
The yellow text is what I put in for naked versions - for male models I also added '1boy' next to the character name
Just don't ask me what all the figures mean
Hopefully attached are zip with what names worked - the title of graphic is the character name I put into prompt.
For those that didn't work I tried using a suitable Lora found on civitai - most of the ones I found were made by WhiteCaliberFox & Midnightcrawler but it was a bit hit & miss into the styling
I'm not sure which path we should go down - which graphics are preferred - pony or nova ?? or is there some other checkpoint that may be suitable ??
I first tried 'using xxx checkpoint & use a random seed until I get a graphic I like & write down the seed' - problem is, if I altered the prompt to add/remove clothing etc... it screwed up the original graphic .... failure
I have today instead used one of the new graphics (sally hi.png) into 'pnginfo' to get a general prompt & copied that into 'txt2img', then using JUST 'ponyDiffusionForAnime' & changing the name in the prompt I get some 'reasonable' graphics - however some characters will not work
You don't have permission to view the spoiler content.
Log in or register now.
I then tried again using 'novaFurryXL_illustriousV9b' & repeated the same thing again - just changing the name - some names worked & some didn't, some names came up with naked version, but even when I added clothes to the text the graphics didn't get too badly corrupted
You don't have permission to view the spoiler content.
Log in or register now.
Anyway, the 'prompt' I used for both checkpoints is:-
You don't have permission to view the spoiler content.
Log in or register now.
Just don't ask me what all the figures mean
You don't have permission to view the spoiler content.
Log in or register now.
Hopefully attached are zip with what names worked - the title of graphic is the character name I put into prompt.
For those that didn't work I tried using a suitable Lora found on civitai - most of the ones I found were made by WhiteCaliberFox & Midnightcrawler but it was a bit hit & miss into the styling
I'm not sure which path we should go down - which graphics are preferred - pony or nova ?? or is there some other checkpoint that may be suitable ??
seronis
Member
- May 9, 2022
- 173
- 302
/me lurks awaiting more random marine images
FoxyTails
Member
- Feb 9, 2020
- 321
- 788
Just a request, when you do, can I ask you also post the prompt/lora/checkpoint/seed etc you used, that way others (ok mainly me) will hopefully be able to replicated them in the same style & maybe even come up with new scenes/clothing etc
Xel101
New Member
- Oct 8, 2021
- 6
- 13
The nova ones look a little to much like ken penders art yet i love the dressWell, I took the bullet today to try & do what I suggested about common AI graphics in earlier post.. Oh dear, it's not as easy as I thought.
I first tried 'using xxx checkpoint & use a random seed until I get a graphic I like & write down the seed' - problem is, if I altered the prompt to add/remove clothing etc... it screwed up the original graphic .... failure
I have today instead used one of the new graphics (sally hi.png) into 'pnginfo' to get a general prompt & copied that into 'txt2img', then using JUST 'ponyDiffusionForAnime' & changing the name in the prompt I get some 'reasonable' graphics - however some characters will not work
You don't have permission to view the spoiler content. Log in or register now.
I then tried again using 'novaFurryXL_illustriousV9b' & repeated the same thing again - just changing the name - some names worked & some didn't, some names came up with naked version, but even when I added clothes to the text the graphics didn't get too badly corrupted
You don't have permission to view the spoiler content. Log in or register now.
Anyway, the 'prompt' I used for both checkpoints is:-
The yellow text is what I put in for naked versions - for male models I also added '1boy' next to the character nameYou don't have permission to view the spoiler content. Log in or register now.
Just don't ask me what all the figures mean
You don't have permission to view the spoiler content. Log in or register now.
Hopefully attached are zip with what names worked - the title of graphic is the character name I put into prompt.
For those that didn't work I tried using a suitable Lora found on civitai - most of the ones I found were made by WhiteCaliberFox & Midnightcrawler but it was a bit hit & miss into the styling
I'm not sure which path we should go down - which graphics are preferred - pony or nova ?? or is there some other checkpoint that may be suitable ??
shirotora84
New Member
- Sep 4, 2019
- 4
- 5
I was thinking 'wax figure', but yeah. She looks rather weird. I definitely prefer the smoother pony model. Maybe because it's closer to the original art style?
Dang man, it's not KP Bad. I mean it's not great but comparing it the Lord of the Uncanny Valley?!
I think "score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_anime, source_questionable" are Pony exclusive tags, because I hardly see them in Illustrious prompts.
Trying to recreate a Rouge picture for example, I used the following prompt (shamelessly sto... adapted from novaFurryXL_illustriousV7b prompts in generated_prompts.txt of the other babysitting game):
anthro,female,adult,rouge_the_bat, giving_blowjob,floating_penis,blue penis,white top, black bra straps, pink shorts,deepthroat, drool, masterpiece, plain white background,full body reference,best quality, detailed fur, detailed face,detailed eyes, beautiful and aesthetic
Negative prompt: blurry,ugly,human
Steps: 40, Sampler: Euler a, Schedule type: Karras, CFG scale: 5, Seed: 2791761429, Size: 1560x1531, Model hash: dca273b7a2, Model: novaFurryXL_illustriousV9b, Denoising strength: 0.6, Version: f2.0.1v1.10.1-previous-224-g90019688
Trying to recreate a Rouge picture for example, I used the following prompt (shamelessly sto... adapted from novaFurryXL_illustriousV7b prompts in generated_prompts.txt of the other babysitting game):
anthro,female,adult,rouge_the_bat, giving_blowjob,floating_penis,blue penis,white top, black bra straps, pink shorts,deepthroat, drool, masterpiece, plain white background,full body reference,best quality, detailed fur, detailed face,detailed eyes, beautiful and aesthetic
Negative prompt: blurry,ugly,human
Steps: 40, Sampler: Euler a, Schedule type: Karras, CFG scale: 5, Seed: 2791761429, Size: 1560x1531, Model hash: dca273b7a2, Model: novaFurryXL_illustriousV9b, Denoising strength: 0.6, Version: f2.0.1v1.10.1-previous-224-g90019688
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
FoxyTails
Member
- Feb 9, 2020
- 321
- 788
Interesting how a slight change can give vastly different results.
As a complete newbie I've also beenstea... borrowing prompts & mucking about trying to get something I like that 'looks' similar to existing graphics (although I've been looking into creating 'missing' scenes mainly).
Not really succeeding repetitively though - sometimes it look too 'cartoonish' while others look too 'human/realistic' or just doesn't look like the character (especially if I try & change camera view) - just like in brutulf examples (incidentally never really mucked about much with the default settings of SD interesting about 'denoising' I assume that's the one in 'hires fix')
Here's (one of) my prompt I've been mucking about with for Tails, although I have no idea what some of it actually meant to do LOL, although resolutions does really muck up some renders, guess it's knowing what size to use I guess for specific scenes
- the white part is the 'default' prompt, the highlighted parts are what I either added/removed.
These are the results I get
Personally I like 'default' as 'modern/new' variation while 'default/no alpha/api' looks more 'cartoonish/old school'
What do others think ?
As a complete newbie I've also been
You don't have permission to view the spoiler content.
Log in or register now.
Not really succeeding repetitively though - sometimes it look too 'cartoonish' while others look too 'human/realistic' or just doesn't look like the character (especially if I try & change camera view) - just like in brutulf examples (incidentally never really mucked about much with the default settings of SD interesting about 'denoising' I assume that's the one in 'hires fix')
Here's (one of) my prompt I've been mucking about with for Tails, although I have no idea what some of it actually meant to do LOL, although resolutions does really muck up some renders, guess it's knowing what size to use I guess for specific scenes
You don't have permission to view the spoiler content.
Log in or register now.
- the white part is the 'default' prompt, the highlighted parts are what I either added/removed.
You don't have permission to view the spoiler content.
Log in or register now.
These are the results I get
Personally I like 'default' as 'modern/new' variation while 'default/no alpha/api' looks more 'cartoonish/old school'
What do others think ?
You don't have permission to view the spoiler content.
Log in or register now.
im partial to the one on the bottom right. default prompt no 'alpha/api'. but in comparison to how cream looks with the updated ai model, i dont know how well it'll mesh. im pretty sure tails was always meant to be younger than sonic so it makes sense to me for him to look small/young.
FoxyTails
Member
- Feb 9, 2020
- 321
- 788
Phew.. been busy. Based on deathspawn246 liking the 'cartoonish' version I thought I'd do a rouge gallery of all the characters in game & see how they all look both 'realistic' and 'cartoonish'
Here's my cartoonish versions...
Please note that for Cosmo I HAD to use a
For Hershey.. well, I did my best trying to describe him/her ... yes 'Her'
- I always though the pizza delivery was a bloke but apparently
Also for Nina.. can't find any results about her on the
A couple of names I've come across often are 'Tikal the Echidna' and 'Wave the Swallow'.
Comments anyone ?
Here's my cartoonish versions...
You don't have permission to view the spoiler content.
Log in or register now.
Please note that for Cosmo I HAD to use a
You must be registered to see the links
For Hershey.. well, I did my best trying to describe him/her ... yes 'Her'
- I always though the pizza delivery was a bloke but apparently
You must be registered to see the links
Also for Nina.. can't find any results about her on the
You must be registered to see the links
. Did find 1 reference to a fan made character for the Sonic universe but no other info - maybe a change to another character perhaps to make thing easier to produce AI Graphics ??A couple of names I've come across often are 'Tikal the Echidna' and 'Wave the Swallow'.
Comments anyone ?
Last edited:
thelastkingdom36
Member
- Jul 16, 2022
- 296
- 483
Wow! The Blaze one is really doing something for me. The stance, the smile, pretty hot not gonna lie. Anyway I like these cartoonish versions. The only couple of nitpicks I have is Sonic is to short., he should be as tall as Knuckles there. And Hershey looks like she want to still my kidney lol
Last edited: