thank you
Tutorial Stable Diffusion AI: Automatic1111 Beginner's Guide
- Thread starter Myexistence
- Start date
Ito yung basic links na inumpisahan ko aside sa mga napag-usapan natin dito sa thread bago ako nag-try paandarin sa potato pc ko he he:
You do not have permission to view the full content of this post. Log in or register now.
Good alternatives yan kung ubos na yung free tier sa Google Colab.
Nakalimutan ko pala itong You do not have permission to view the full content of this post. Log in or register now. ay ok din for testing dahil maraming choices sa models. Binalikan ko lang para may choices ka rin at yung iba na interisado sa AI Art. Sa mga UI testing lang yung hilig ko kasi, sorry. Saka ko lang binabalikan yan pag may nagtanong. Marami kasing branches ang AI na sinusuyod ko, isa lang ito na di ko talaga area. Mas naaakit ako sa benefits ng AI sa knowledge kaysa Arts kaya sorry kung maliit yung mabigay ko sayong support. Try mo yung Stable Horde option, meron yang working web link at may software sila for the same purpose to generate AI Art online using SD - yung Lucid Creations GUI to be exact using Godot engine. Mabagal pero acceptable yung results at napakaraming models to choose kasama yung paborito mo he he, Loras, atbp. Baka alam din ni Myexistence yan. Magandang back-up yan dahil halos unlimited. Mag-iintay ka lang ng turn mo. Sa palagay ko, sulit na ito at yung provided ni Myexistence. Wala ka ng hahanapin pa he he. Inulit ko yan kanina para makumpara silang lahat. I-ready mo lang email/s mo ha! Good luck.
Nag-testing ako ng SDA1111 (full featured) portable version ngayon na cpu lang yung gamit. Ayos ang takbo, stable at hindi nag-hang pc ko. Medyo matagal mag-generate pero nakakapag-browse ako ng normal. Pag tumakbo na sa Win 10, nagdadagdag siya ng mga 4-5 GB at 2GB sa processing - max total ram sa pc umabot sa ~11GB kasama yung ibang processes. Mas safe siguro sa cpu-mode is 16GB. check nyo dito. Isa-isang pic ko lang ginagamit dahil di pwedeng pwersahin.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
lods bakit iba UI ng sakin lite version
Maraming forks yang Stable Diffusion na nilabas ng Stability AI, pero yung Automatic1111 ang mas preferred ng users. Depende yan sa version din ng pinagkuhaan mo ng installer. Ang latest version ng SDA1111 ay v1.4.0.
Kung hindi yon ang na-install mo, hindi nga magkakaparehas yung GUI nya. Same din kung sa Colab ka napa-andar. May gumagawa ng lite versions para doon salow specs or base sa unas labas at sak lang nagdagdag afterwards. Yung NMKD, EasyDiffusion are examples bukod pa doon sa cpu-versions.
Yung iba kasi compact na may hidden dropdown selection kaya di mo makita. Check mo muna at napakarami nyang versions sa net. Ikaw lang makakatuklas nyan he he. Ang importante magamit mo muna or else install A1111 para sunod ka sa latest features.
ano po gamit nyo sa colab lods? saka meron ako nabasa optimized SD ba yun pwd kaya yun dito
processor: Intel(R) Core(TM) i5-8265U CPU @ 1.60GHz 1.80 GHz Installed ram: 4.00 GB (3.85 GB usable)system type: 64-bit operating system, x64-based processoredition: windows 11 proos build: 22621.1848intel core i5 8th gen nvidia asus laptop FHD 1080 X409FL - FT561T
Ako? Yung nasa first post ni MyExistence at mga nightly builds, pero di na ako nag-Colab ngayon. Interisado lang ako noong una due to curiosity para masanay, pero di na ngayon.
Yung SD cpu-mode lang ang pakay ko ngayon to train how to use other AI on cpu-mode too. Di ako interisado sa graphic generation. Daan lang yan for other projects. GPU-extensive kasi AI apps, nahihirapan akong gawin offline habang nag-iintay ng budget for better hardware.
Sa specs mo, sinabi ko na sa other thread yung masasabi ko. You can try at your own risk sa cpu-mode. Good at napalitaw mo, pero na-try mong mag-generate? Estimate ko, mga 5-10 minutes ang generation per pic, by default settings (512x512, 20 sampling steps, etc. assuming you don't run out of memory. Pag nag-BSOD ka, you'll have to be ready for your recovery tools he he. Nakaka-corrupt ng windows files yan. Ako, ilang beses na bumigay ang pc sa ganitong gawain sa kakapiga ng hardware ko. Be warned!
Ang suggestion ko kung gusto mo talagang gamitin yung SD cpu-mode ang recommended ng marami sa Reddit is 16GB or more para safe ka sa cpu-mode. (Yung sa akin 12 GB, pero kumain na ng +2GB sa start ng OS sa dami ng apps ko, tapos another +4GB sa initialization ng SD, plus another +4GB sa processing ng 1 image kaya pigang-piga. Pag-isipan mo. Pero baka matipid yang nagamit mo he he. Good luck.)
Note: lalakas ang kain nyan depended rin sa laki ng checkpoints na gamit mo kaya be warned at kailangan paisa-isa lang ang genration or lower the pic size.
Bigay mo nga sakin yung link ng SD mo, para subukan ko sa akin.
Ang pinakamatipid daw ayon sa mga developers is SD Openvino at itong SD NCNN, Na-test ko pa lang yung huli pero di stable sa +512x512 size, napakabilis nga siya pero sa 256x256 lang. yung Openvino , baka this week ko subukan kung ma-hanap ko kiliti sa installation. Pero base yan sa pinakaunang mga SD. Yung GUI pic mo is enough to generate good images. Pwede naman hiwalay yung Controlnet at iba pa.
Last edited:
shishoumaru
Honorary Poster
pwede din kaya yung unstable diffusuion?
pang generate kaso need dapat paulit ulit yung muka paano po kaya promp nun same face?
pang generate kaso need dapat paulit ulit yung muka paano po kaya promp nun same face?
Unstable diffusion is just the same any SD fork. Based yan sa "Automatic 1111" na tinawag nilang fast-stable diffusion.
Kung simple lang na editing pics while reataining faces, you can use simple img2img or Instruct Pix2Pix, inpainting from text, editing image from text, face swapping using controlnet (using "masking" method"), or train SD AI using a database of your face in Dreambooth. Hanapin mo sa net yung mga guides nyan from simple to hard ayon sa order ng binigay kong methods. Naka-highlight na yung search words mo para guides. Kung alam mo yang You do not have permission to view the full content of this post. Log in or register now., malamang miyembro ka ng Disc@rd dahil meron silang guides doon para sa hinahanap mo: You do not have permission to view the full content of this post. Log in or register now.
Sa first 3, ala-tsamba ka sa prompt and weights para mapasunod mo yung limited capability ng AI using mask prompt, prompts, strength, image guidance, guidance, checkpoints, even seed. Trial and error yan. Medyo mas pino yung results using Controlnet, pero and best is using training SD AI using as many pictures of you in Dreambooth. Less stressfull sa paggamit ng prompts dahil yung mukha mo yung priority ng AI.
Kung simple lang na editing pics while reataining faces, you can use simple img2img or Instruct Pix2Pix, inpainting from text, editing image from text, face swapping using controlnet (using "masking" method"), or train SD AI using a database of your face in Dreambooth. Hanapin mo sa net yung mga guides nyan from simple to hard ayon sa order ng binigay kong methods. Naka-highlight na yung search words mo para guides. Kung alam mo yang You do not have permission to view the full content of this post. Log in or register now., malamang miyembro ka ng Disc@rd dahil meron silang guides doon para sa hinahanap mo: You do not have permission to view the full content of this post. Log in or register now.
Sa first 3, ala-tsamba ka sa prompt and weights para mapasunod mo yung limited capability ng AI using mask prompt, prompts, strength, image guidance, guidance, checkpoints, even seed. Trial and error yan. Medyo mas pino yung results using Controlnet, pero and best is using training SD AI using as many pictures of you in Dreambooth. Less stressfull sa paggamit ng prompts dahil yung mukha mo yung priority ng AI.
Naghanap ako ng magandang method kaysa sa dati kong nakagawian. Ito yung best sa lahat.
1. You do not have permission to view the full content of this post. Log in or register now.
2. You do not have permission to view the full content of this post. Log in or register now.
Last edited:
jhnnlasnts
Forum Veteran
Pag nag-eenter na ng prompt, may lumalabas na connection error kaya di makagenerate
Sa Colab? Baka mahina internet mo.
DetroitGuru
Forum Veteran
Nice, keep it up ts
jhnnlasnts
Forum Veteran
Mabilis naman sya... Ngayon hindi ko na maaccess ulit, ayaw na magreconnect or even install
May limit kasi paggamit nyan sa Google Cloud. Pinapahiram ka lang ng hardware online to run SD. Magpalit ka ng email, baka ma-reset kung sakali.
thanks for sharing idol bat kaya disconnect ng disconnect sakin
Maraming dahilan ng disconnection (depende kung anong error pa yung lumabas dyan), pero maaaring galing sa poor internet connection ninyo or may ginagawa yang Google sa gumagamit ng A1111 ngayon. Di lang kayo ang may complain na ganyan. Maraming fix yan kaya hanapin ninyo sa google using this prompt query "google colab frequent disconnection fix" and try each if they work. Yung iba sa paperspace or runpod nila pinapaandar yung notebook if for some reason you are forced to use the notebook method. Marami namang guides sa net to do it. May free at ρáíd din.
Kung walang umubra, hanap na lang kayo ng free online service na merong decent parameters similar to A1111: You do not have permission to view the full content of this post. Log in or register now.
It's all up to you.
Similar threads
-
-
UltimateAI - AI Enhanced WordPress Plugin with SaaS for Content, Code, Chat, and Image Generation
- Started by bappii
- Replies: 9
-
-
-
Course Udemy | Intro to Generative AI for Business Professionals | Limited Time Only | April 08, 2024
- Started by Luwiiiii
- Replies: 0
Popular Tags
about
ai
ai detector
alternative
ang
anong
app
artificial
artificial intelligence
best
bot
chat
chat gpt
chatgpt
chatgpt 4
content
course
free
generate
generator
google
gpt
here
image
in
intelligence
midjourney
need
openai
or
part
pasabit
prompt
prompts
quillbot
resources
text
that
tools
trending
unlimited
useful
using
video
website
websites
with
writing
you
your