Midjourney V5.2 Course (Hyper Realistic Cinematic AI Photography & Image to Video with Runway Gen 2)

in this video I'm bringing you a practical course to create stunning hyper realistic cinematic AI photos using ultimate prompt structure on mid-journey and how to bring your mid-journey Creations to life using the new frontier of cinematic storytelling one-way Gen 2 I will show you the full workflow to create a cinematic trailer with consistent characters using face swap methods step by step this includes upscaling your mid-journey images using topaz gigapixel AI creating videos from your images with Runway Gen 2 and the optimal settings for upscaling Runway videos to 4K using topaz video AI I will also show you how to use New Motion setting and Camera controls in Runway as well as free alternatives for these tools let's start with ultimate hyper realistic photography and cinematic prompt structures there are many different ways to approach the creation of an AI generated image and prompting Atomic prompting is one of the methodologies for organizing a prompt that helps make the life of creators easier and their workflows much more efficient Atomic prompting is inspired by Nature with that we will organize components of our prompt according to four layers organism molecules atoms and parameters which together builds the image it's a simplified and structured approach to prompting organism level is the most foundational decisions we need to take for an AI image generation here we will decide on the art direction we will take and describe the scene in the simplest way possible do we want more realism or more painting style who is subject and what is the action in the scene photography type is critical to notify mid-journey for our desired creative Direction and its first ever decision we need to take for example do we want to have a street or architecture photography do we want to have a fashion photo shoot or product photography what if we want to create cinematic AI photography this element is one of the most important parts of our prompt structure after that we are describing our photo and finding answers for who is the subject in photo what are they doing you don't have to answer all these questions but I recommend to include these critical details to maximize your creative control on the final outcome and we know from my early prompt studies that mid-journey gives it most high ranking to tokens mentioned early in the prompt that's why these elements are in the organism layer when we break down elements in the organism layer we are reaching to molecular layer represented by yellow color chord and they're about secondary details of subject what is their location and what's their outfit look like in which year our subjects are living here we have also camera controls camera name and photographer style it's highly valuable to describe short type in the prompt look how Vibe and emotion changes when we apply different camera angles to the same subject after this we are going to the atomic level where we start breaking down molecular level elements to further details for example details of fashion and outfit description of background and more Atomic level can be considered as advanced prompting and it's not necessary for everyone unless you really want to control every detail in your image for a majority of people organism and molecular level will be more than enough last decision we need to take is deciding on right parameters for our image for example you if we want to apply default mid-journey Aesthetics to our photo and if yes how strongly I made videos about this before so you can refresh your memory if you want to understand stylized parameter better you can also use new work parameter together with stylize to create really unique looking unconventional images just keep in mind to use same value for both parameters for balanced results if you don't want to apply default mid-journey Aesthetics which includes an artsy touch then just add style raw parameter to your prompt it's safe to say that style raw defaults to photorealism slightly better than stylize and not his hostile Rob pays attention to the original prompt more closely and people follow me for a long time nosdad and not the biggest fan of combining style raw with stylized elements since I believe they have different purposes but I don't want to limit your creativity so go try to combine them if you prefer a mix style caming said that in the prompts which includes historical costumes or generally in fashion photography mixing up stylized with raw makes significant impact I think this is due to the training data for mid-journey's default Aesthetics style is being heavily influenced by fashion photos then finalize your prompt with your desired aspect ratio when it comes to creating cinematic images only thing you need to do is adding word cinematic to transform your prompt the keyword cinematic is important as it prompts mid-journey to interpret the image as a cinematic photo taken from an actual movie scene in the molecular level we are using a combination of shot types camera names with cinematic features filmmakers with unique art style and we are setting the tone with here emotion and lighting together this will provide amazing value if you have something specific in mind for cinematic prompts we have costume and set design in the atomic level we don't have to go that deep unless we really want to control every detail in our image if you love to keep things simple just use the organism level together with minimum amount of parameters for a simple but effective cinematic looking image if you prefer to create your own Cinematic Universe use this prompt as an example I created the movie from an alternate universe where Batman vs Joker was a Bollywood movie from India I recommend using the style raw parameter to fully leverage the visual Style Elements we incorporate in the prompt especially if you mention a camera name or director's name with a unique style this will exclude Standard mid-journey Style Elements and make the algorithm truly focus on your style keywords since we learned how to create auto realistic cinematic AI photos with mid-journey let's apply everything we learned in practice and bring our images to life to create a cinematic trailer in my trailer I also wanted to create consistent characters which appears throughout the trailer not just random people I will use combination of tools to bring those images to life and some of them are actually paid tools but I will give you budget alternatives to achieve similar results for some of the tools without spending too much money first we will decide on what kind of AI field we want to create it's required to have some type of direction to start building our world I was interested in the mystery of mass sighting of Sky events that occurred in the sky in Fatima Portugal in 1917. this is based on true events that happened in Portugal with hundreds of people witnessing a glowing orb moving fast in the sky and people were shocked by it now I will create a fictional film synopsis about this story create characters and start building my universe using chat GPT and mid-journey I use this prompt here is the peace chat GPT Rod now let's continue with creating our Cinematic Universe here we will use mid journey and also use chat GPT as prompt assistant I will use this prompt for chat GPT to help me create my trailer it creates the trailer for me seen by seeing with opening shot introduction of the main and secondary characters conflict tension Climax and more this is helpful but somehow I need to translate this to a visual language of atomic prompting mid-journey can understand for this job I'm using this prompt then chat GPT translates my trailer scenes to a table with visual language that mid-journey can understand plus I have the table I will use chat GPT as prompt the system to build my mid-journey prompts based on elements on the table here are the mid-journey prompts created by chat GPT for visuals of my trailer now I will go ahead and copy and paste these prompts to Mid journey to generate scenes for my trailer you can of course adapt and change them a bit character names don't mean anything for Mid Journey so instead of that you can describe your character's physical features we will change our character's face in the next step for consistency additional to that I will probably need more shots of landscape people and architecture of Fatima of Portugal there is a quick way to build your cinematic World in mid-journey use this prompt this prompt will bring all the necessary elements for my trailer and combining it with the chaos parameter will ensure that I have all of these elements in my grid then I can upscale a lot of these elements let's say landscape and then use strong variation feature of mid-journey to create different landscape shots for the same time with same team for my trailer I need to create my characters Beatrice and Dr Harrison and use their faces consistently throughout the different scenes in my trailer I saw some people use following prompt structure in combination with impainting for character consistency this prompt creates a great view with multiple angle shots of the same character you can also use strong variation button for more similar shots use vary region button for in-painting mode and select extension of one of the images in the grid while selecting be sure you include a tiny part of your original image and change prompt to your desired pulse or C and here is the result I think this method is great for illustrative drawing style of characters for realistic images I found this method to be less flexible I will show you a more flexible method called face swapping first we will start by designing our main protagonist using this prompt many are satisfied with the look of your character upscale it for creating consistency we will use face swapper add-on for Discord face swapper is a cool bot on Discord allows you to add any face to your mid-journey generations to install this add-on you need to create your own Discord server which is free and really simple once you create your server go the link in description and install inside face swap bot to your Discord server after accepting invite and adding this bot to your server you need to also add mid Journey bought I designed Beatrice's face using mid-journey and now I need to save it to face swap board in order to do that go to your server and type slash save ID this command will need a name for your character and you will need to upload her photo here so later we can use it in our new generations after hitting enter I can start generating new scenes for Beatrice in this scene for example Beatrice is in train station and her face looks like someone else to fix this I will right click on my image head over to apps and click on in swapper remember this spot will copy paste last face you save to the new scene you created and here my character's face is placed in New Body posture for another scene this method is much easier to create lot of scenes with consistent looking characters we can change our character's face expression as well to do that click on vary region button to switch to impainting mod pick lasso tool and select your character's face add emotion or face expression to The Prompt when you hit enter mid Journey will generate same character with a different emotional state once you upscale a lot of them apply face swapper again and create consistent looking character images with different emotions here we made her very happy additional to face swapping in painting comes quite handy for consistent character creation for example we can inject our character's face directly to the scenes to do that upload your character's face to Discord right click and get the link to the image now create your scene that you want your character to be after upscaling click on vary region and select the face of your subject now we need to change the face come to the prompt section and copy paste link to your original character's face with that we created an image prompt together with text hit Arrow your original character design is injected to the new scene since we have our scenes with consistent characters created we can upscale those images unfortunately mid-journey upscaler alone isn't enough for us to truly bring out the details and make our images pop we need another tool topaz gigapixel AI is my go to upscaler although it's a paid tool the results are worth the investment one of the standard features of topaz gigapixel AI is its ability to upscale images in box swiftly this is especially handy when you have a series of images that need to be enhanced getting started is simple install and open top as gigapixel AI once you've got the app installed on your Mac launch it add your images you can drag and drop entire folders individual images or use the browse option to locate and select your images on the right menu you'll find an option to set the upscale ratio I personally prefer to upscale to 4X choosing the AI model there are various AI models available from my experience the standard model gives the best results however if you're unsure the app has an Auto mod which selects the optimal model for your image for settings like suppressed noise remove blur and compression I usually leave them in auto mode this way topaz's AI can make the best decision for the image mid journey is already quite good at handling faces so I don't usually enable face recovery or gamma correction however if I'm working with a scene that has multiple faces especially in a crowd I might turn on the face recovery option once you're satisfied with your settings click on save image if you've added multiple images top has gigapixel AI will upscale all of them in bulk we created our images using mid-journey and upscaled them now it's time to create our video footages for the cinematic trailer using Runway Gen 2 and upscale these videos using topaz video AI one way is a versatile tool with both free and paid versions the free version does come with a watermark but it's a great way to test the waters if you're on a budget consider using picker Labs on Discord as an alternative to Runway and capcat video upscaler is an alternative to Topaz video AI you can create your video footage from a single image or you can also type a prompt in addition to the image you uploaded giving more context to Runway you can drag and drop your mid-journey images or click on upload a file once you've uploaded your image you'll need to adjust settings if you're using a free account some options might be restricted both Runway and topaz video AI have interpolation option I recommend to check one ways into interpolation this means small transition between frames using a machine learning algorithm I tested interpolation of topaz video AI against runways and realized for faces Runway interpolation does a better job however if you have no visible faces in the scene top has video AI brings great fluidity to scene I recommend to uncheck Runway upscaler because it's pretty slow we can use external tools like topaz video AI for this job Runway recently added new settings called motion motion allows you to set intensity or fluidity in your video higher values result in more motion default value is 5. custom camera controls are the latest feature from Runway it allows you to set the motion of camera you can choose to pan horizontally or vertically you can zoom out or zoom into your subject one of the advantages of Runway is its ability to generate multiple videos simultaneously this allows you to produce several videos and select the best one instead of waiting for hours initially videos will be 4 seconds long but you can extend them up to 18 seconds I observed shots from behind tend to maintain coherence much better now let's export our Runway Gen 2 videos to Topaz video AI for upscaling using plus button here I will show you my choice of settings on top as video AI one by one let's start from the top I want to upscale this video to high resolution so I will choose 4X here and I keep here as 24 frames per second we don't need AI stabilization because movement is actually an element increasing realism and it's something we want to have we also don't need AI motion Tabler frame interpolation is optional if you already use Runway interpolation in your video then you don't need topaz interpolation for enhancement let's set the video type to Progressive and for AI model I will pick Proteus we don't need any extra grain or noise in the image down below we have some encoder presets I like to choose prores and is called the guy will choose 422 HQ hit export with this settings you will get much better resolution for your videos take a look at video outputs adding voices of our characters to the trailer brings them to life for this job I will use 11 laps first I need to choose right voices for my characters 11 Nets allows you to sample voices so you can find the best possible option if you click on speech synthesis you will find some pre-selected voice options if you go to Voice Library you can find even more options and filter depending on age gender accent etc for my trailer I need few sentences to be voice over so I can use them between my scenes in the trailer I decided that voice name Charlotte fits best to my main character Beatrice I am copy pasting Lions chat GPT created emptying generates for music I use epidemic sound because I already had a count there but you can use pixabay for free stock music to bring everything together I use cap cut desktop version capcat's desktop version is much more capable and intuitive than web version here you can upload all the runway videos you created and upscaled using import button then when you hover over the file you will see a plus button this will allow you to quickly add the video footage to your timeline also you can add text if effects transitions automatic captions stickers and more if you love to apply color correction for visual consistency between your Clips click on the footage head over to adjustments on the video details window scroll down adjustments and you will see detailed color and light related settings for cinematic look another important detail you may need is to decrease or increase music levels depending on the scene for this job head over to your music file in the timeline on the corner of the media details window you will find this square button called add keyframe this button allows you to add keyframes to your music so you can adjust volume levels scene by scene their mobile app has a feature called 3D Zoom you can upload some of your mid-journey images and create this cool Parallax effect now let me show you the final trailer a light in the sky [Music] in the sky I'm investigating Miracles around the world something happened there that day in Fatima Vatican wants me to find the truth they want to keep people in dark and dumb so they would never know the truth a group within the government doesn't want this to be known Portuguese government is worried about implications of this event [Music] you need to be careful you expect me to believe in miracles I'm a scientist and I need to stay in the boundaries of science I need a proof Beatrice a scientific evidence I will solve this with or without your help Professor I owe this to people of Fatima [Music] [Music] subconscious is like a recording of everything that has happened in your lifetime in the background we need a way to unlock those memories if we want to solve this puzzle [Music] I believe these people we can solve this together [Music] I believe everyone can create art that's why I'm creating these high quality mid-journey courses for free so everyone can watch and learn without paying hundreds of dollars hopefully this video was truly helpful for you if so give it a thumbs up and subscribe to my channel for more awesome tutorials if you want to learn more about creating AI photos videos and everything AI related click here

Channel: Cyberjungle

Views: 64,489

Keywords: midjourney, midjourney tutorial, midjourney ai tutorial, runway gen 2, topaz photo ai, topaz video enhance ai, midjourney prompts, topaz gigapixel ai, capcut, stable diffusion, leonardo ai, prompt engineering, midjourney face swap, midjourney consistent character, how to use midjourney, runwayml, runway ai, runway gen2, runway ml, pika labs, pika labs tutorial, eleven labs tutorial, pika ai, pika labs image to video, gen 2 runway, midjourney inpainting

Id: nHSVaWmuQnQ

Length: 19min 26sec (1166 seconds)

Published: Sun Sep 17 2023