/v/ - Video Games

Vidya Gaems

Index Catalog Archive Bottom Refresh
Options
Subject
Message

Max message length: 12000

files

Max file size: 32.00 MB

Total max file size: 50.00 MB

Max files: 5

Supported file types: GIF, JPG, PNG, WebM, OGG, and more

E-mail
Password

(used to delete files and posts)

Misc

Remember to follow the Rules

The backup domains are located at 8chan.se and 8chan.cc. TOR access can be found here, or you can access the TOR portal from the clearnet at Redchannit 3.0.

RIP David Lynch

/vhs/ David Lynch Marathon


8chan.moe is a hobby project with no affiliation whatsoever to the administration of any other "8chan" site, past or present.

Reminder that 8chan.se exists, and feel free to check out our friends at: Animanga ES, Traditional Games, Comics, Anime, Weekly Shonen Jump, /b/ but with /v/ elements Official 8chan server: mumble.8ch.moe:64738

(1.40 MB 2160x3840 Smug Lain Spoiler.jpg)

(823.19 KB 1360x768 Rosa gazes into the abyss.png)

(842.38 KB 768x1296 Marisa x Reimu.png)

(1.16 MB 1360x768 Venom Onesie.png)

(772.19 KB 1616x2880 Vivian (Oblivion Classes Art).jpg)

AI General ( Stable Diffusion / DALL-E / Chatbot / Voice Cloning ) Anonymous 01/12/2025 (Sun) 20:15:47 Id: 2fee5a No. 1060375
Smug Spoiler Edition [Copied from previous thread] ── ✦ ✦ ✦ ── Use this thread as a catch-all for AI related topics. Try to stick to vidya & help your fellow anons out. PLEASE SHARE EVERY NEW RESOURCE YOU FIND OR CREATE ── ✦ ✦ ✦ ── Resources: https://pastebin.com/fLMeAMb1 Previous Thread: https://archive.fo/ZMDOK Previous threads are moved to >>>/ais/ [500 posts or Page 13 before next thread]
>>1072634 That's one fine looking Vita, AI has come a long way in a short time
(425.48 KB 653x490 1611961930716.png)

>>1072634 >Thanks. I like this art style a lot. It's great. I encourage you to do more because i will save everything you post here or on /h/. Thanks, anon.
>>1072436 Those are some cute Chens
(207.61 KB 1024x1024 73K9M7KB2PVAMS17TF6RDRN2C0.jpeg)

(205.59 KB 1024x1024 NM5G03TDHTYYVGN753NQS8QZQ0.jpeg)

(216.25 KB 1024x1024 SR1E143AGWKSN5XAF5Q086SBK0.jpeg)

(215.41 KB 1024x1024 FM19D692PYNE07D5GG1A79B2X0.jpeg)

(199.23 KB 1024x1024 RHTMK7C6ECXTAMJFKZ4Q5MPWA0.jpeg)

https://civitai.com/models/1236928/il-fallout-talking-head-style?modelVersionId=1393991 Fallout 1/2 Talking Head style for Illustrious. My last conversion, so totally new LoRAs start tomorrow.
(443.57 KB 2243x750 CHARs.jpg)

Speaking of mugshots. I didn't get any new big images done because I was generating character thumbnails, Been using the styles drop-down in reForge Stable Diffusion WebUI for character prompts, but it was getting very cluttered. If I edited the CSV file it uses in another application it stops recognizing the file & wipes everything so I couldn't even alphabetize it. I finally had enough & searched the extensions page for something better. I found the Stylez extension does what I'm after & could import all the junk I already had in the existing styles drop-down menu. Not only does it alphabetize, but also uses separate folders for categories, can use thumbnails & it has a search function. So I had 30 character prompts that are mostly recognized by the BetterDays checkpoint with a few exceptions. Now in the WebUI I have these 30 character thumbnails I can click on to insert prompts for these characters which is nicer than the cluttered drop-down I was using before. I think I used (artist:gerph:0.5) & flat colors tags for the art style. Prompted to do a simple gray background, but the AI has some wildly different ideas about what that means between gens. I was considering zipping all the JSON files & thumbnails for these character prompts to share here, but I don't think my prompts are good enough on all of them. Some still need work. I dunno if anyone else uses that extension. It's not made for ComfyUI AFAIK so only works on A111 & its derivatives like reForge.
(1.63 MB 3840x2160 Chel Slide.jpg)

(1.70 MB 3840x2160 Edward P.jpg)

>>1072639 It's fun to mess with when I have the time.
>>1072931 How are these such high quality?
>>1072933 im also wondering this, >>1072931 do you upscale them after genning them? do you natively gen them as jpg at that res?
>>1072933 Time, effort, photoshop, inpainting, fixing a lot of minor things. Styles with complex shading are harder to clean up. That Disney animation style is simpler, with no shading & some sketchy bits / hatching here & there, so it can be a little rough & still look good, but I still try to clean it up a lot. >>1072939 These are far from native gens. I do a lot with them & it takes a while. I've mentioned in past threads, after final upscale I crop a portion, usually the face, & inpaint to make it as clean as possible, then place that cropped portion back into the full image. With a better GPU I wouldn't need to do it that way, but regardless, faces usually have thinner / cleaner lines because I've done that. I don't always do it. Sometimes inpainting the 1st upscale is enough. I do 2 upscales. 1st is hi-res upscale in TXT2IMG, then a ton of inpainting, then the 2nd final upscale in IMG2IMG tab using SD Upscale. May alternatively do a regular upscale in Extras tab when SD Upscale is doing stupid shit & I don't feel like taking the time to combine its successes to mitigate its failures as each tiled upscale is time-consuming. I fine-tune a lot in Photoshop. Fuck with color balance, vibrance / saturation, curves, apply minor sharpening & soft overlay filters to doll them up. I add grain & a subtle halftone pattern to dither gradients & obscure artifacts. Very minor chromatic aberration, not overkill. I tweak skintones, but my display is not a good representation of what other people are seeing so hopefully skintones look decent. Could go into specific photoshop actions I use, but dunno how many people use it for this & I've heard actions are fickle about PS versions anyway.
First new LoRA in a while: Amy Rose outfit. Was once curious if this could be done with tags or a general Amy LoRA alone due to seeing something else I don't remember. It was not so it went on my maybe list. Had some free time so I decided to crank it out. Vivan was actually going to be today, but I brought her forward to account for >>1071307 so the then newest work was slotted into her former place. I'm curious how popular this will be since it's from media the average person will actually recognize. https://civitai.com/models/1251451/il-amy-rose-outfit Big project wound up with over 200 images, over half of which are now tagged (it's easier to tag characters than styles due to consistency in tags. Just copy-paste a list of outfit tags, clear the non-applicable, and add a few for the background/poses.). Hope I can throw it into training tomorrow and have it release on 25th.
(2.17 MB 3840x2160 Amy Begs.jpg)

(2.26 MB 3840x2160 Amy Begs No-TXT.jpg)

Amy Rose? Artstyle is "Myriad - Illus x Niji" https://civitai.com/models/1226171
>>1072957 >Time, effort, photoshop, inpainting, fixing a lot of minor things. Sounds like a job, you should find a way to make bank out of that.
>>1073059 >pics sorry to say amy, but sonic's only into human woman
(958.79 KB 2126x2160 Boring Fast.jpg)

>>1073059 Nice that there's no character overlap so a simple crop can get you this reaction image.
>fly the slimey skies! Sally form Mortimer and the Riddles of the Medallion. I wonder if anyone here has played it. https://civitai.com/models/1239537/il-sally-mortimer-and-the-riddles-of-the-medallion?modelVersionId=1396980
(122.39 KB 540x661 Sonic outrun.jpg)

>>1073129 It does seem to be the most consistent part of his tastes over the years.
(1.71 MB 3840x2160 Vivian Kiss Me I'm Irish.jpg)

A drunk Vivian for Saint Patrick's Day which is next month on the 17th.
>>1073243 Needs more purple.
>>1073251 >>1073243 Agreed. Off the top of my head the hat could be striped with alternating green and purple sections since it's already segmented. The part with the ribbon on it should be purple for contrast. Her shirt could have purple buttons. And the skirt could be green and purple tartan.
>>1073243 I want to impregnate that drunk Vivian.
>>1073251 >>1073255 Did full green outfit for Saint Patricks. Was also going to gen a green tophat (like leprechauns wear), but that wasn't working so went with a paddy cap. Goal was to go really irish with it & also have her thick ass spilling over the edge of the table. AI wouldn't cooperate enough to add more, but I think what I managed is adequate. >>1073261 Same.
>>1073263 Its good but it doesn't feel like Vivian without at least a little purple somewhere on her outfit imo, even if it is st patrick's day
(1.66 MB 1536x1536 bunny tea 01.jpg)

(1.76 MB 1536x1536 bunny tea 02.jpg)

(1.66 MB 1536x1536 bunny tea 03.jpg)

(672.74 KB 640x640 aaaAAAAAAA.png)

I am a epic noob to this and I've been trying to figure this out all night. I am on a fresh install of the Automatic1111 webui and got a bunch of stuff from Civit. So I'm off to the races, right? Well I try to make shit and it ranges from pretty alright with noticeable fuckups to monstrous blobs. I've read a bunch of stuff from the pastebin and such and I'm fairly confident I've not fucked up installing or anything. If I copy the generation data and such from a picture on Civit I can sometimes get it to a workable state, or flat out copy the picture using the seed. But other times some stuff doesn't work out. I'm trying Pony or Illustrious, with and without LoRAs, high and low Guidance, whatever... I am not expecting stuff like 2fee5a's things because he said he spends a lot of time editing stuff, but I don't know how to just get to a state where I get a mostly normal picture. What's like, a good "blank slate" model and settings that won't fuck shit up? Or am I being too picky, and is this just how it is?
(1.71 MB 3840x2160 Vivian Kiss Me I'm Purple.jpg)

(1.71 MB 3840x2160 Vivian Kiss Me I'm Purple 2.jpg)

(1.85 MB 2160x1440 Cover Art 2.png)

>>1073251 >>1073269 >>1073255 I did the slow quick edit of making darker greens purple. Also did the even more slow quick edit of striping the hat which started to remind me of a train conductor's hat. After seeing purple versions, I still prefer full green for the holiday theme. IMO Viv doesn't require purple to be Viv. In lewds she's naked with no purple in sight. She usually, but not always, includes purple in my gens. For example, I like her in the Anonymous-style outfit: >>1070038 TL;DR Dressing & undressing Vivian is fun. I'll post the PSG-style cops again because it's neat & another anon made a nice logo for it. >>1073290 I'm not on that PC right now to look at my settings, but I can at least tell you even if you're going full autism with inpainting, upscaling & edits - the more complex an art style the more difficult it is to get it looking good. Realism is the hardest as the mind knows perfectly what "real" things look like. If you go in a more cartoony or sketchy direction there's more room for the mind to fill in the blanks so "looking decent" is far less strict.
>>1073298 If anyone can figure out what the typefaces are for the new P&S title card, I can do another one using that format, too.
(710.62 KB 2048x2048 native gen hi-res upscale.jpg)

(694.52 KB 2048x2048 inpainted.jpg)

(950.39 KB 2048x2048 photoshopped.jpg)

>>1073290 Here's an long post detailing my methodology. It's good to have a starter prompt ready to go with a handful of tags that you know improve the output then just add to it. I've made an effort to show what I'm using & how I used it. Prompt: best anatomy, detailed eyes, large pupils, detailed pupils, thick eyelashes, shiny skin, detailed background, chiaroscuro lighting, incredible lighting, detailed shadows, vintage dining room, glossy wooden table, dark green wall, teapot, carrot on plate, steaming tea, 1girl, teen, skinny, medium breasts, pale skin, green eyes, very long hair, straight hair, swept bangs, long sidelocks, black hair, bunny girl, (anthro:0.6), freckles, freckles on shoulders, freckles on breasts, human nose, small nose, long black rabbit ears, strapless white dress, frilly dress, baggy clothing, bare shoulders, half-closed eyes, closed smile, soft blush, one eyebrow raised, looking at viewer, sitting at table, one hand lifting tea cup, drinking tea, elegant, beautiful, (traditional media, oil painting, classic media), Negative prompt: IL_Neg, body paint, watermark, heart pupils, (small head), (2girls, group, multiple_girls), depth of field, out of focus, blurry, cuffs, (double ears, human ears:1.5), (hairband, white rabbit ears), tea on plate, When making bunny girls I often struggle to remove human ears on this model. I had (anthro:0.6) in prompt to see if that helped & (double ears, human ears:1.5) in negative. I believe the nose is more red because anthro is in prompt although it's at a lower weight. Steps: 35, Sampler: Euler a, Schedule type: Align Your Steps, CFG scale: 4, Seed: 1550528052, Size: 1024x1024, Model hash: 05369ba2dd, Model: Better_Days_Illustrious_XL_V01, VAE hash: 235745af8d, VAE: sdxl-vae-fp16-fix.safetensors, Denoising strength: 0.4, Clip skip: 2, Hires schedule type: Align Your Steps, Hires upscale: 2, Hires steps: 15, Hires upscaler: ANIME_4x_IllustrationJaNai_V1_135k, I use hires upscaler at a denoise of 0.3 - 0.4 typically. With a simpler style it can potentially go higher, with more realism it generally has to go lower. I don't always use IL_Neg embedding in negative prompt, but it can help anatomy issues. Checkpoint: https://civitai.com/models/623891?modelVersionId=1143094 VAE: https://civitai.com/models/140686 Upscaler: https://openmodeldb.info/models/4x-IllustrationJaNai-V1-ESRGAN IL_Neg embedding: https://civitai.com/models/993446/illustrious-embeddings Embedding link 404s. Tried alternative, but civit.ai is 404ing that too. I'll just upload the embedding I used: https://d.uguu.se/jwJDBglB.safetensors Note: Embedding didn't work for me unless I included --disable-safe-unpickle in launch settings. Inpainting: Masked content: original, Mask blur: 15, Inpaint area: Only masked, Soft inpainting enabled, 35 steps, CFG: 4, Seed: random. <Face - 0.48 denoising strength <Cup+Steam - 0.45 denoising strength Avoided inpainting hands since it kept fucking them up, Sometimes it decides to be a bitch about that. Face & steam benefited from inpainting. Other areas didn't benefit much so I didn't bother. Didn't want to make this one look like it took a stupid amount of effort. Photoshop: >Filter - Blur - Surface Blur: Radius 3 Threshold 4 Loses detail, but unfortunately necessary. Inpainted areas have noticeably different grain that must be cleared first. Not needed in all cases. Could inpaint all the thick grain away or else prompt it away. But in this case, I have to wipe the grain in PS & reintroduce it to make it evenly grained. >Duplicate layer >Filter - Sharpen 40% - Sharpen filter on the duplicate layer, set layer to 40% Opacity, merge with original layer >Action: "grain" 1 - slight at 40% Fill, merge with the layer below This re-introduces grain evenly across the image. Dithers gradients, hides artifacts, film grain is useful. <!! Kept this current layer for later. Continued on duplicate layer. >Duplicate layer >Action: Vibrant glow 15% Opacity >Action: Sunglow 2 15% Fill >Image - Adjustments - Color Balance: Shadows -3 -2 +3, Midtones +4 -1 +7, Highlights -4 +6 +3 (color tweaking) >Image - Adjustments - Vibrance: -15 +5 (makes skintones paler) >Curves: input output, 0 3, 82 77, 181 179, 255 250 (brightened areas that became too dark) <!! All these changes are merged with the layer from earlier at 60% Fill >Duplicated layer, named new layer FX (Action only applies to layer named FX) >Action: CA Diagonal 1PX >Action: Fade Edges (only affects layer named "Final effect", I fade edges of this layer to avoid lines on the edges) >Action: Hallftone Overlay 1 I sometimes increase the scale of grain or halftone if they appear "too fine" for the resolution. Vibrant glow & Sunglow 2 are from "Kubota Artistic Pak I" found online. Other actions I either created or don't remember where I found 'em. Could've done more Photoshop, but the section is too long already. I'm using Photoshop CC 2019. Heard actions can be fickle about versions so my actions may not work for you. Photoshop Actions: https://f.uguu.se/tvSHcswC.7z (the Styles are also necessary for some actions) The simpler answer is use a model that easily generates good outputs with low effort & work from there. If you get something like the base NoobAI XL model you'll be tard-wrangling it for ages to get anything good. BetterDays by contrast "just werks" with a decent default style so you have a good starting point to work with. Artist tags, style tags & LoRAs can take you the rest of the way.
Tried out trying to make >>1069826 in a realistic style. I have no idea what I'm doing when it comes to making realistic people and the artifacts from the art style of the training data likely make it a poor choice for realistic images.
Following Sally, here is Sarah Anderson, a significantly less (but still fairly) obscure 90s Lucasarts girl. Unfortunately I've yet to find a checkpoint other than base Illustrious 0.1 that doesn't hallucinate a bunch of detail on her boots and collar area. https://civitai.com/models/1240241/il-sarah-anderson-lucasarts-outlaws?modelVersionId=1397800 Also I released a model early on tensor art because it needed a lot of testing/gallery population before prime release due to combining both a 3D option (I have no idea what makes good 3D in Illustrious so that's a lot of trial and error) and a double digit number of costumes. Might as well blow the 50 credits that expire at the end of the day on that rather than buzz which carries over https://tensor.art/models/832724507480311865
(2.22 MB 3840x2160 Milk Thief Cirno.jpg)

(1.36 MB 3840x2160 Misty Dream.jpg)

(322.04 KB audio.wav)

(726.04 KB opensource.wav)

>>1073496 Open source voice cloning? Interesting.
(316.04 KB dunmerline.wav)

(316.04 KB poshdunmer.wav)

(655.04 KB wutdunmer.wav)

Still can't do Dunmer voice. Hybrid is actually worse.
>>1073496 So the idea I mention in the previous thread is now possible: >>>/ais/5110 Only took a few minutes gathering audio for it, getting a decent output & throwing it together in Audacity. Found ffmpeg script to combine image with audio, but still had to cut the end off in xmedia recode as it went silent after 13 seconds. Wouldn't mind a local install if there were a nice WebUI & it's not hard to get working. Crazy times we're living in.


Forms
Delete
Report
Quick Reply