Is this an AI drawing?? Is this an AI drawing? If I install an AI program, can I create it too? When I searched on Google, it is said that the WEBUI integration pack (AI picture making tool) is made by a competent person and it is easy to install. I will fix the link to the installation description . Click next to install, unzip the WEBUI Integration Pack and move all the tar files into the unzipped WEBUI folder. Right-click to install. Run Powershell. Find Powershell with Windows + Q button and right-click Run as administrator. Enter Set-ExecutionPolicy -ExecutionPolicy RemoteSigned (there is a comment field) Enter Y and close Powershell . It proceeds...Comfortable integrated pack Automatically proceeds after Y enter...Convenient...Is this an integrated pack? Wait until the Internet window automatically pops up. Installation is complete . I clicked Create. Fortunately, it was a safe (?) picture . To give a rough explanation, it's a level that I haven't gone to, so I'll try to create 4 boxes for writing prompts (commands I want) and writing negatives (commands I don't like) ... Resolution 768X512 4 sheets 3080TI Time to make... Graphic card 100% used ( 3080TI standard 350W) 4 female characters in suits created The more you write the prompt box, the more you write the desired picture Probability up test If you add glasses to the prompt, a picture with glasses comes out We are picky eaters, we want delicious pictures If you don't know what it is, replay what you've done for 2 days in order Press F5 to refresh Chrome and it will return to the initial settings Stable Diffusion checkpoint on the top left Stable Diffusion checkpoint = Model selection space What 's the most important thing when dating a girlfriend? That's right... It's a picture style Stable Diffusion Checkpoint = Model (picture style) Use the famous Orange Mix model on the bulletin board If it's your first time, let's try it because the Orange Mix model is not installed. After downloading the orange mix model from the hugging face site, turn off stable diffusion and put the downloaded orange mix model in the folder of the video. And if you run the WEBUI, you can select the orange mix model . WEBUI update is also convenient if you click Check for Function Updates... The integrated pack has been made very convenient by a good person. I entered the click edit and entered the --no-half-vae command, and the picture creation failure symptom disappeared (I didn't know this, so I retried several times...) I gave the command only once and double-clicked it from the next time. WEBUI is a graphics card resource. 100% is used while creating a picture Witnessed using 11.5G of 3080TI 12G video memory Power consumption is 350W Recorded Ready Now let's go back to the main computer... If you've made it this far, everything is ready . ) In the input video, only sound pictures should come out... If you put nsfw in the negative box, they say that naughty things are prohibited... When I tested it, there were times when it suddenly popped out after ignoring it... The thrilling world of YouTube (it becomes an ineligible channel...) Any picture is created? = No jam It's a game whose purpose is to create the picture I want Saro Vertical option = The higher the resolution setting, the more power of the graphics card is required Number of batches = Number of pictures created when creating 1 click I created 4 1920 1080 with 3080ti.. .It takes about 1 minute GPU usage 100% 350W consumption... Now let's do it Initialize once and prompt column = write what you want Negative column = write what you don't like When you enter the Novell AI site, the tag page AI listens well to commands. Gaze composition, direction, limbs, behavior, facial expression, clothing, hair color... etc. (Is this 2023 Princess Maker?) Alas... You have to put in a negative nsfw to film on YouTube... (I'm sorry too ...) So, if you make 4 pictures in 1girl ideal state, 4 pictures will come out at your discretion. I'll make a viewpoint first. Insert a focus tag. From now on, a picture focusing on the upper body comes out . Now specify the background. An upper body + city background picture appears . It is possible to customize almost infinitely... I can skip this . Input upper body + city + blonde hair + ponytail + frontal gaze Frontal gaze Erase and look back Input Upper body + city + blonde hair + ponytail + look back Infinite tastes... Up to wing tail (?) Possible (?) Unfortunately, we cannot show you. .. I don't like looking back at the striped shirt as a costume . Delete it ... Come back with a frontal look. Remove the shirt. It's set up so that only the upper body comes out... What are gym shorts? I deleted the upper body option and made it visible even the bottom. YouTube is dangerous because it keeps coming out... Let's put on a suit. Upper and lower body + cafe + blonde hair + ponytail + suit + frontal gaze. Glasses here? Upper Body + Cafe + Blonde + Ponytail + Suit + Frontal Gaze + Glasses Upper Body + Cafe + Blonde + Ponytail + Suit + Frontal Gaze Upper body + Café + Black hair + Ponytail + Suit + Frontal gaze + Raise arms Because it's for YouTube, Upper body + Skinless suit Fixed option... Unfortunately, I was going to film this far and finish it, but a new WEBUI function called Control Net This came out... Awesome... I tried the ControlNet installation process . The default state is ControlNet menu No menu Enter url in the extension option (address in the comments) Install When restarting the installation, close all WEBUI and controlnet data on the Huckingface site Download (address is in the comments) Download complete Put the controlnet data in the WEBUI installation folder path WEBUI start Controlnet menu creation complete Controlnet installation explanation End Move to main computer (3080ti) Testcom is 1080ti...slow The pictures created so far are automatically included when you open the folder . Turn on the option to enable inserting pose photos in the net box. What if you want to import a pose for a figure photo? Select an open pose for both Let 's make a female character who imitates the figure pose Controlnet open pose option creates figure pose wire frame Possible , but
the basic function has a low recognition rate, so ControlNet needs to be used here too. Image 2 Image tab also uses ControlNet open pose . Possible to create a skeleton and modify its location (created right before the end of the shooting...Mongamonga function continues to exist) Possible to create a skeleton without a photo To recognize a skeleton photo Processor = None in ControlNet Model = Open Pose PoseMaker Skeleton Ready to use Skeleton If you have it, you can immediately input large movements and start, making it easier. If you put a command in the prompt, you can change all hairstyles, clothes, and facial expressions. Mysterious open pose function There are many control net functions (personally, open pose function is the best). I use it when I come Image 2 Image Control Net Enter an ai picture made by someone else If you turn off the control net function and create it, mongamon is a pity... Turn on the control net and select canny canny and create it? Line drawing with controlnet canny function + picture drawing with quotation marks image 2 image Copy much neater than basic function this time, use controlnet hed function When I put the ponytail command, a ponytail was created on the picture. The AI picture was copied by AI and the AI picture was created . Copy of replication of replication... Color change is easily possible All functions that cannot be shown in the video... A week will be short... Everything becomes a game if you enjoy it... But what is copyright for AI pictures?
