>How
Download the package and extract it on some fast drive that have good amount free space (100gb at least). Then
----Run 1stAIdBag_install.bat
--------Install ComfyUI and Manager. This is kinda bare minimum you need to install. You can install workflow's missing node via manager. Some user inputs needed during this.
--------Check if your ComfyUI runs after this. If it does....proceed
--------Install Custom nodes. This will install all custom nodes needed in my workflows, also other workflow's commonly used nodes (this might take some time)
--------Install Attentions. Highly recommend this as they give speed boost for generations.
--------Download essential WAN files and one of the WAN GGUF models (Q8, Q5 or Q2) Better your rig, bigger Q to get
------------NOTE: Downloading these models take time as those are several gigs
--------OPTIONAL: Download QWEN, GIT, or SDXL stuff if wanna play with image workflows

Run comfyui. On the first run it will take some time to load as some nodes like reactor will download some models needed. Once loaded, open the latest version of the workflow, re-select the all the models, vae, clip and loras. This is something you need to do every time you open up a new workflow, so the files point into locations in your computer. See the vid: https://screenrec.com/share/mkgZDHG5uI

After that, check other settings, type in prompt and hit run.

Buy me a beer: https://ko-fi.com/1staidbag

>Already got old install?
----Backup (cut) models and user folders somewhere
----del old ComfyUI_windows_portable folder. Then do as guided above
----paste model and user folders back into your fresh ComfyUI folder after install

>What do i need?
-Windows, RTX card, pref at least 8GB of VRAM, 32GB of RAM, drivers up2date
-You can run stuff with lower rig specs but then you would need lower models, lower resolution, lower clip duration, etc
-There are some info/help to run stuff with ATI cards in prev threads
-NOTE1: Major change on the pack, nothing is pre-installed, all nodes are getting installed via 1stAIdbag.bat now. Pack is now much smaller, 7megs

>If/when you need help....
A) take a screenshot of your whole workflow (and log console window if possible)
B) state what workflow are you using (i'm not going to try fix other's workflows for you)
C) what specs on your rig

>Where to get loras?
https://civitai.com/search/models?baseModel=Wan%20Video&baseModel=Wan%20Video%202.2%20I2V-A14B&baseModel=Wan%20Video%202.2%20T2V-A14B&baseModel=Wan%20Video%2014B%20t2v&modelType=LORA&sortBy=models_v9%3AcreatedAt%3Adesc
(you need to make account and enable show adult content)
https://civarchive.com/search?type=LORA&base_model=Wan+Video+14B+i2v+480p&is_nsfw=true&sort=newest&page=2
(old 2.1 loras that got removed from civitai due to their nsfw bullshit policy)

Old thread(s)
https://archived.moe/r/search/subject/ComfyUI%20portable%20for%20newbies/

Up with ye

Thx for the fix, got it running with attentions installed. What is flash attention?

Thx a lot man

File: 1758235891120793.jpg (76.9 KB)

76.9 KB JPG

what the fucking point to add guide in a request channel? just help people requests or shut up. people still spam bump. so you are definitly useless

>>20360636
some of us have requested to learn how to make them ourselves, so it fits in the request area...

>>20359903
flash attention is kinda like older version of sage. (not really as they work differently). But i incl. it as some older cards don't support sage 2.x. Attentions are like optimizations that take effect when you hit that run button, they split the image noise into smaller blocks so its more vram/ram friendly and also faster...in simple terms

>>20360817
>not trying to do something too fancy
>10 second gens
>720p

yeah....no. Maybe 4-5sec at 480p @ 4steps with lighning loras and Q5 model....and time

anyone know where to get fetish loras like Piss?

>>20360919
yeah I figured, got that from the older threads in the archive, it's why I deleted my post
probably just going add another 16gb of ram soon, even if I don't really understand how that improves the process, it takes less time to gen probably?
If my 4-5 second vids turn out alright and I feel like I get the hang of it I'll just bite the bullet and get myself a new gpu as well

>>20361169
hugging face

Bumps for bag

been trying IAMCCS new FLF2V nodes, but I have a hard time getting nice movements out of it, like with PainterFLF2V nodes.

At least he seems to be actively developping it.

>>20362075
I did try those during the weekend, I got ok motion but loads of colorshifts....if the movement is bad just in the beginning then make sure that first clip is set to first segment (skip previous latent)

File: ruined cumshot ext 02280000.mp4 (3.6 MB)

3.6 MB MP4

ive searched civitai and hugging face. cannot find a lora to accomplish the ruined orgasm. this one is from motionmuse and i cant get anything to look like it.
any ideas?

>>20363238
im pretty sure that is penisplay (pworship) lora. I can't remember it that has cum baked in. gonna do some tests and let ya know

>>20362442
For the colorshift, he released a fix yesterday. But that maybe the thing that broke motion since your workflow with the previous node still works fine.

Been tryint to install, unfortenelly huggingface_hub seems not fixed. someway to go around this?

>>20363654
did you do a clean install? and using the intaller on op post? also what card you have

>>20363664
>https://files.catbox.moe/rbwu4v.rar
already noticed that i installed v3! taking care of v4 now! thanks!

>>20363680
that issue should be fixed on v4....but DO FRESH install, don't try to overwrite it over old one (backup your models folder somewhere if you already downloaded them)

>>20363700
already got it running...

whats the best way for face swap or generating images from photos without getting to "ai look".

Used simple reactor for face swap, and SDLX for generating... but the faces get weird.

Sorry, i am really new in this!

>>20363955

>>20351930

File: 03-03_22-00-15_00001.mp4 (2.5 MB)

2.5 MB MP4

Does anyone know how to avoid the mini dick after the deep throat? Been trying with other prompts, dick loras, different seed but nothing seems to help
Workflow is SVI

>>20365634
try to prompt at the last segment "at the end woman moves her head away" or "woman licks sucks the tip of the penis" or load some handjob lora and un-load the blowjob one in the last part....

wan can only operate on stuff it can see in the first frame, in svi, you kinda need a bit of luck how the last segment will end (it's last frame will be next segment first one)

do NOT use "woman pulls penis out from her mouth".......or you can, for shit and giggles =). Actually there is lora for penis pullouts, i dunno how good it works on blowjobs

>>20365692
Thanks, good stuff. Seems picky but all logical when you think of it

>>20365723
on svi, prompting is bit tricky...as you have to kinda guess how the scene end in order to prompt the next one....and then there is that luck, some seed might give you perfect results, other some funny wtf moments

>>20363610
ty. also trying to find the "implied bj" one where it shows the back of the womans head. hard to replicate...

File: 15798110936164705_00001(1).mp4 (3.5 MB)

3.5 MB MP4

>>20365857
i think the first part is sensual_bj lora rather that pworship (has no cum trained in).....cum part.... tried cumv2 with prompting it oozing rather than shooting......work in progress

Help with missing nodes, I am trying to use the 1stAidbag Anime to real but anytime I try to install the custom nodes ReActor and Layer style I get an error. does anyone have a fix or workaround? I found an Alt for the layer filter but nothing for the face swap and boost

>>20366150
...the installer didn't install Reactor itself?

Because Reactor is a bitch to install. Check your terminal to confirm that the issue is related to insightface (likely) and try the following:

https://github.com/Gourieff/ComfyUI-ReActor?tab=readme-ov-file#troubleshooting

>>20366159
It did not but I just tried to re run install custom nodes and it broke my Comfy so I am going to have to do a clean install anyways so hopefully it will work this time (my 6th time!!!)

>>20366150
reactor is pre-installed and nsfw-patched from installer, so it should work....go you your manager, view custom nodes, and select filter "import failed" click on that and give me some more detail (if you got multiple ones failed list; like pulID, instantID.....then your insightface is not installed)

>>20366175
install comfy, then nodes, then attentions (installs) insight, If reactor still fails, install attentions again....if still fail, gimme some terminal info

>>20366175
>>20366181
Reinstall worked and now Reactor is working, I'm surprised I made it all week without noticing something was wrong.

>>20358908
new to this, is there a particular sage that is better? i have a 5090.

i installed v4, but now my video previews don't automatically play, they are paused. i don't see anywhere in the setting to change it back to how it was before

What do I do about add ons rar? Do I just put it in when I install everything else?

What is the best Lora to create realistic facials and what preset to use ? I tried many with different Loras, settings and prompts but I can never get rid of that digital rendering layer that breaks all immersion.

>>20366565
Got good results with
https://civitai.com/models/1922973/f4c3spl4sh-cumshot-i2v-wan-22-video-lora-k3nk
and this kind of prompt (keep only what you need)
>f4c3spl4sh, Tthe woman starts receiving a cumshot on her face from the penis. The force of the ejaculation is strong. The cumshot splashes on her faces and drips on her breasts as she seems to enjoy having lots of cum on her face. The womans eyes are open as she receives the semen. The camera is positioned close to the woman's face and breasts, capturing the moment of male ejaculation directly onto her face and breasts. The semen is seen forcefully erupting from the penis and landing on her skin, creating a visible pool of semen on her chest.
this gives that kind of result >>20362915

>>20366562
Installer and extract that when needed

>>20366469
Leave it auto. There is also sage3 for 50xx, I dunno if Wan can benefit from it

>>20366475
Do you mean on video combine node, or previews on the left side (assets)

File: 03-04_14-30-22_00001.mp4 (3 MB)

3 MB MP4

>>20365692
>>20366822
Thanks boss for the help, final result is nice

>>20366822
*installer will extract it when needed, you don't need to touch it

>>20366748
Thanks, nice results indeed. I need to work on that basis, obviously the results could be really upgraded with high definition input, Q8 and nothing less than 720p output

>>20366845
Yeah, Q8 and 720p are the limit for my setup, but I can't really use the worstation in parallel anymore if I do that.
Working on a prompt/lora sequence that works and then I'll fire some gens during the night.
Still on the fence for the TripleSampler (on or off)

I can't for the life of me get a consistent flashing set up. just want ladies to take off their tops and show off their tiddies. I can use loras to do all kinds of shit but not a simple flash.

pls send help

File: Screenshot 2026-03-04 100958.png (253.4 KB)

253.4 KB PNG

>>20366822
i mean the previews on the left side.

also, anyone know why this happens? never happened before v4 for me (but i also never used sage until v4 so idk if it is related to that)

>>20366923
Search shirtlift in civitarchive

>>20367212
Does it Gen the videos ok thou? I never seen that error.... I would think it's related to comfyui frontend... But error is titled so it points to clip error

>>20367221
when that error occurs, the video does not generate. i have had it happen twice since i started using v4/sage. i usually have a queue going, so i only notice when i check back and see that it is no longer generating videos.

this particular time, it made about 20 videos before this error happened

>>20367221
it might be related to this?

https://github.com/Comfy-Org/ComfyUI/discussions/12699

some people said a fix is Launching ComfyUI with --disable-dynamic-vram is the workaround.

I a dumb, could someone tell me how to do that lol?

>>20367261
Open comfyui GPU sage bat file with your notepad... Add that line at the the end

>>20367273
thank you, i know it was a very computer-illiterate question

>>20367273
run_nvidia_gpu_SageAttention.bat right?

>>20367284
Yep

>>20367287
like this? or before pause?

sorry for the dumb questions lol

File: Screenshot 2026-03-04 110223.png (16.8 KB)

16.8 KB PNG

>>20367297
ugh lol forgot to attach

>>20367261
This might help or not..... I've never quoue 20 gens. But could be after each video you make models don't get unloaded completely or cache get cleaned. So after while ram/vram gets filled with this residue "junk" and you basicly run out of usable memory until you reboot comfyui

>>20367299
Before pause..... After or before that sage part

>>20367303
i have been using v3 for a month or 2 and have literally 1-200 in queue regularly with no problems, so i am guessing is it that problem, assuming comfyui updated when i deleted everything and installed v4

File: Screenshot 2026-03-04 110751.png (15.9 KB)

15.9 KB PNG

>>20367311
like that? i just want to make sure i am doing it right so i can better troubleshoot if it does not fix it

>>20367317
That is correct now

>>20367312
Yep comfyui has got quite few updates since v3

>>20367329
yeah and that one just happened a few days ago. that is the only thing google shows me really, when trying to find anything related to "VBAR"

the funny thing is, i had it running overnight, and when i woke up, i knew something was wrong, bc my room was much colder than normal, from it not running haha

fingers crossed

>>20367329
hypothetically, if i cannot get the problem fixed, would going back and doing a fresh install of v3 roll comfyui back? i am guessing not and that it just gets the newest version whenever you run your install bat

>>20367351
V3 installs latest comfy as well

>>20367506
ok so it did not fix the problem. i posted it on github and they said

We found and fixed an issue and it is now likely resolved for you if you update comfy and refresh your requirements.txt

so new dumb question: how do i do that?

>>20367982
i read that you need to run update_comfyui.bat by doing that just gives the message "The system cannot find the path specified."

>>20367999
there are no update bats in my pack....i removed them because 99% of time people would break their comfyui when using it.....always better to backup models folder somewhere and do fresh install

update bat updates comfyui sure.....but it also might change crucial elements (listed in requirements.txt) like torch and cuda version --> attentions will not work. AKA you will most likely break more major stuff just to fix minor issue

>>20368012
ok yeah i figured if nothing else, i will just do another fresh install. hopefully their fix works

>>20368012
oh also, is the point of vacking up the models so you dont have to get them again from your install?

like should i copy them back after it installs comfy, but before installing custom nodes, attentions, and the skip installing the wan and gguf since theyre already backed up?

>>20368029
(i know you explain this in your posy, paste model and user folders back into your fresh ComfyUI folder after install, i just don't know if "after install" means just the first step in installing comfyui, or if you mean after all of the steps are done)

>>20368029
once comfy installed.....but before you do all that that i think there might be easier way to fix your issue (just found the github thread)

just update the aimdo......seems like that is the lib causing the issue, gpu memory leak

close comfy
go into comfyui/python_embedded/ folder
right click and run console
type
>pip install comfy-aimdo==0.2.6
boot comfyui up

File: Screenshot 2026-03-04 155814.png (225 KB)

225 KB PNG

>>20368040
you might need to explain this like i am a 5 year old lol

i right clicked in this folder, open in terminal, and typed it in and got this

>>20368072
try:
>python.exe -m pip install comfy-aimdo==0.2.6

>>20368082

Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Apps > Advanced app settings > App execution aliases.

should i just do a fresh install lol?

File: Screenshot 2026-03-04 161409.png (96.2 KB)

96.2 KB PNG

>>20368087
ok i did the app setting thing then got this

PS D:\rbwu4v\ComfyUI_apprentice_portable\python_embeded> python.exe -m pip install comfy-aimdo==0.2.6
python.exe : The term 'python.exe' is not recognized as the name of a cmdlet, function, script file, or operable
program. Check the spelling of the name, or if a path was included, verify that the path is correct and try again.
At line:1 char:1
+ python.exe -m pip install comfy-aimdo==0.2.6
+ ~~~~~~~~~~
+ CategoryInfo : ObjectNotFound: (python.exe:String) [], CommandNotFoundException
+ FullyQualifiedErrorId : CommandNotFoundException


Suggestion [3,General]: The command python.exe was not found, but does exist in the current location. Windows PowerShell does not load commands from the current location by default. If you trust this command, instead type: ".\python.exe". See "get-help about_Command_Precedence" for more details.

so i did that and i think it worked

>>20368115
yep...that should be now updated (and fixed your issues) hopefully

>>20368132
yep after i do a bunch of gens, i'll let you know it it fixed it in case anyone else has the problem

also, bumping my comment that the left side previews don't play automatically like they did for me before v4, in case you know why

if i am just doing several videos, with different images but no change in any other settings, prompt, lora etc, can i right click the vram-cleanup?

i think i read that somewhere before but idk

>>20368012
One thing I did, was modify comfyui to pull all ./model/* folders from a workspace level folder ( ./shared/ ). This is up and outside of my ComfyUI instance. I did the same with inputs/outputs. Now I can spin up a new instance of comfy to mess with if I want to go down a dark path of PIP hell. Or start a clean instance from your pack (./ComfyUI_apprentice_portable_V4/). Life is much happier now.
Working on pushing the instances to docker images now, with Runtime, Lora, and big models all their own images for easy updating.

>>20368142
i don't think they never have autoplay'd?

......consider you queued 20 clips, come back, click assets tab open, and all those 20 vids starts auto playing at the same time.......i don't wan't to even consider it. How hard is it to play play button?

>>20368155
you can save all your models into different drive if you want and use the tool from add-ons folder Easy-Models-Linker.bat

File: 2026-03-05 023618.png (289.8 KB)

289.8 KB PNG

>>20368177
mine previews have never autoplayed....and thats good

File: 238554.jpg (43.3 KB)

43.3 KB JPG

>>20368180

>>20368177
weird, mine always autoplayed in v3, but it would pause any that weren't on screen so i guess it wasn't an issue

idk why it did autoplay for me before

>>20368238
maybe you had some custom node installed that enables that...or maybe it's browser setting...but still don't why would you want something like that

>>20368177
Totally appreciate what you built. It got me into it. I'm still a newbie with the workflows.

I've made something focused on the install/deployment. I hope to share soon.

It's a single ps1 that fully bootstraps everything.

It iterates workflows for missing models and downloads the models (or allows you to skip, also it allows you to cancel and save the portion you already downloaded, then resume)

It then patches the workflow with the model location ( I started using Lora Manager, the folder structure got complicated ). It also has an alias file for old models that are no longer named that way, allowing you to match what are in some workflows, with other models.

Also has E2E testing of all workflows, to make sure they load without error and run (via the api).

Give me some time, i'll post a minimalist version of it. Right now, it's a pretty big repository as a whole. And i'm about to add in Docker support so I can spin up my instances in the cloud or other machines easily.

>>20368364
glad to hear i inspired someone who knows what he does ;) i suck at ps.....seems like you got loads in the oven

make it a good one so i can retire from this project

>>20368412
Thanks Codex 5.3.
I just made a new workspace with it, and we've been iterating all day. I know generally what I want, but it's killing it.

Hi 1st.Ald.bag, one quick question, I have a 5060 and I’m interested in having a workflow img2img just for nudifying, what is the best recommendation to get started with those, I don’t what to spend 5 min waiting for videos to be generated

bump

>>20368942
Try flux klein

>>20368412

are you really going to retire ??