Best Way To Get Consistent Characters In Midjourney! - Portraits, Realistic

one way to get consistent character portraits is to use a split camera method let's use a prompt a woman tasking and dark hair brown eyes natural lighting Kodak portra 160 split into multiple images shot from multiple angos the key part of the prompt being split into multiple images and shot from different angles this gives us a base character portrait collage to work with of a couple of different images showing the same person from different perspectives in the rest of the video I'll cover getting more viewpoints customizing environments changing clothing and more so without prompt asking for multiple images we get a split image showing the same consistent character from different angles in an image collage to get more viewpoints we can take advantage of the pan feature in mid-journey click on the pan arrows beneath the image to extend it I'll pan to the left to get more shots of the same character the new generated images will use the original character portrait collage as a reference this is my preferred method of getting more viewpoints alternatively you can use a custom zoom function to achieve the same effect click on custom Zoom I'll leave the zoom Factor at 2. and we get an outpainted version of the original image collage with more views I prefer using the pan feature because it preserves the resolution of the original image and it also tends to just generate higher quality results than using custom Zoom getting our character in different environments is simple first turn on remix mode in the settings menu we'll keep using the panning pinning feature for this select one of the pan arrows and the prompt appears for us to customize since we have remix mode turned on in the prompt I'll add a location of in a forest after a woman generate the image and our character has been moved to a different environment outdoors in a forest and the facial features are preserved it might take a couple re-rolls to get results you're satisfied with using the same exact method you can put the character in many different places like the beach the city or a cold winter environment for different clothing items we'll use the same exact pan method select the pan arrow and in the pop-up prompt ask for different articles of clothing I'll go with a denim jacket this time I like the the way that the brown one turned out we can get many other types of attire like a bucket hat luxury clothes a red dress or a leather jacket we can also mix and match different articles of clothing and environments like asking for a sundress in the field Outdoors these images all preserve the identity of our original character and allows us a lot of customization options different actions can also be generated using the pan feature let's edit the prompt and ask for our character reading a book holding a pizza playing guitar or playing chess here we do expose some of the limitations of my journey it is inconsistent when generating different types of actions especially with the hands alternative to the pan feature to get custom environments and clothing we can also use the very region in painting tool select very region and in the pop-up interface click and drag to select a section 10 paint make sure to leave a few reference images to maintain the original woman's facial features in the prompt we can ask for a different environment like in a forest or go to the beach a city and try a new outfit I prefer using the pan feature for customizing environments and changing clothing because you get more images but the impending tool will also work let's see if our method also works on different subjects we'll go with the man this time we can put him near a waterfall go to a jungle or visit London let's give him a new leather jacket I think he looks like he might be a chess player You may wish to extend the borders of some images since they're pretty narrow for example here I've cropped out a part of my image collage and I want to extend the environment under edges using now painting we can use clip drop for this purpose it's free to use although you can pay for some extra benefits check the description for links first create a for your cap I'll sign in with Google then head to the young crop tool next just drag your image onto the tool and adjust the borders I'll shrink them a little bit since I don't want to extend my image by that much select next to extend since we're using a free version there will be a cue however this usually goes by pretty fast I've seen long cues up to 400 people but the wait time goes by quickly in a minute or so once it's finished our image gets AI extended borders similar to photoshop's generative feel clip drop uses stable diffusion for this I need to give credit to the post that inspired this video via chaseling on Twitter so go check out this page now at this point you may be wondering what to do about the resolution of the generated images since mid-journey is limited in the resolution it can generate go ahead and check out my video on upscaling mid-journey images and make sure to subscribe to stay updated on the latest mid-journey news and updates sure [Music]

Channel: Tao Prompts

Views: 11,523

Keywords:

Id: kYsmTnxkUNE

Length: 4min 53sec (293 seconds)

Published: Thu Sep 14 2023