Using AI to Make Composites : Firefly, MidJourney & Generative Fill in Photoshop Beta
TLDRIn this insightful video, the creator shares his experience using AI to generate composite images of NFL football players. He details the process of retouching skin with Vato AI, creating backgrounds with Firefly and MidJourney, and refining them with Photoshop's Generative Fill. The artist also delves into lighting techniques used on set, from clamshell setups to edgy action shots with strategic use of softboxes and octagonal lights. He emphasizes the importance of blending AI-generated elements with high-quality photography for a cohesive final image, showcasing the potential of AI in enhancing creative workflows without compromising artistic vision.
Takeaways
- 🎨 The video discusses the use of AI in generating composite images of NFL football players, focusing on the process from photography to post-production.
- 📸 The speaker shares personal history with photography and sports, leading to a career photographing athletes for trading cards.
- 💡 For lighting, the speaker used the Ellen Chrome mobile app to control multiple lights, creating various setups for different shots.
- 🌆 AI tools like Firefly and MidJourney were used to generate backgrounds, while generative fill in Photoshop Beta was used to fix these backgrounds.
- 🖼️ The process of retouching the skin was done using Vato AI, which was also used for batch processing and minor retouching without overdoing it.
- 🌟 The importance of lighting in creating mood and focus in the images was emphasized, with different setups for clamshell lighting and edgy action shots.
- 📈 The AI-generated images were upscaled using Photoshop's Super Zoom feature to improve their resolution for use in composites.
- 🔍 The use of neural filters in Photoshop for harmonizing color and contrast between the foreground and background was highlighted.
- 🌈 Creative techniques like adding a cyan overlay and grain to unify the image and create a more realistic composite were discussed.
- ✨ Catch lights in the eyes were manually enhanced to draw attention and add realism to the final composite image.
- ⏱️ The speaker found that using AI tools like Firefly and MidJourney was faster and provided more options than generative fill for creating backgrounds.
Q & A
What is the main focus of the video?
-The video focuses on how AI is used to generate composite images of NFL football players, covering the use of various AI tools for retouching, generating backgrounds, and blending the images in Photoshop.
Which AI tools were used for retouching the skin of the players?
-Vato AI was used for retouching the skin, to remove blemishes, brighten dark circles under the eyes, and reduce wrinkles.
How did the speaker use Firefly and MidJourney to generate backgrounds?
-The speaker used Firefly for creating heavily blurred images of cityscapes and landscapes, while MidJourney was used for generating stadium scenes and tunnels with cinematic lighting.
What is Generative Fill used for in this context?
-Generative Fill is used to remove the player from the generated background image in Photoshop Beta, allowing the speaker to replace the background with a retouched image.
How did the speaker use lighting during the photoshoot?
-The speaker used various lighting setups, including clamshell lighting, strip softboxes, and a hair light, to create different looks for the athletes, all controlled through the Ellen Chrome mobile app.
What was the purpose of using a black curtain for some shots?
-The black curtain was used to create a contrasting background for the athletes, allowing for a more dramatic and edgy look in the composite images.
Why did the speaker choose to use a round light source for the action shots?
-The round light source was chosen to avoid creating an octagonal catch light in the players' helmets, aiming for a more realistic stadium light effect.
How did the speaker ensure the AI-generated backgrounds matched the foreground images?
-The speaker used the 'Harmonize' feature in Photoshop's Neural Filters to match the color and contrast of the AI-generated backgrounds with the retouched foreground images.
What technique was used to refine the edges of the hair in the selection?
-The speaker used a technique involving painting along the edges of the hair with a white brush on an Overlay blending mode to darken translucent parts and improve the selection.
Why did the speaker not use Generative Fill to create the backgrounds?
-The speaker found that Generative Fill took too long to generate the desired background image and did not provide as many options or as quick results as using Firefly and MidJourney.
How did the speaker deliver the raw files to the client?
-The speaker delivered the raw files almost in real-time to the client, who then made their own selections for their specific purposes.
What was the final step to polish the image and draw the viewer's eye to the player's face?
-The final step was to add a layer called 'catch lights' to paint and brighten the catch lights in the player's eyes, and then use an exposure adjustment layer to darken everything in the shot except for the player's face.
Outlines
🎨 AI-Generated Composite Images for NFL Players
The speaker shares their experience using AI to create composite images of NFL football players. They discuss the process of retouching skin with vato AI, generating backgrounds with Firefly and mid-journey, and using generative fill to fix backgrounds. Additionally, they talk about blending the images in Photoshop and lighting techniques used on set. The speaker's journey from collecting trading cards to photographing athletes for trading cards is also shared, highlighting the evolution of their career and the excitement of photographing multiple athletes in a single day.
🖥️ Post-Production and AI Tools for Image Enhancement
The video script details the post-production process, starting with delivering raw files to the client and then selecting personal favorites for portfolio purposes. The speaker processed the images in Capture One with a high-contrast base characteristic for a punchy look. They used ivato Ai for batch retouching to remove blemishes, brighten dark circles, and reduce wrinkles without overdoing it. Adobe Firefly and mid-journey were employed to generate background plates, with Firefly being favored for cityscapes and mid-journey for stadium scenes. The speaker also explains their keyword strategies for generating specific background images.
📸 Photoshop Techniques for Image Integration
The paragraph explains how the speaker used Photoshop to integrate the retouched player images with the AI-generated backgrounds. They used generative fill to remove the original player from the background and then merged layers for editing flexibility. The speaker also discusses upscaling the AI-generated images using the super zoom feature in Photoshop's neural filters for better resolution. After integrating the player into the background, they refined the mask, harmonized colors and contrasts between the layers, and added finishing touches like blue edge lights and grain for a unified look.
🖌️ Final Touches and Creative Color Adjustments
The speaker describes the final steps in enhancing the composite images. They added a blue edge light effect by painting with a cyan color on a new layer and adjusting blending modes and opacity for a natural look. Grain was added to unify the image and create a cohesive texture across the background and foreground. Additional color unification was achieved by overlaying a subtle cyan color. To draw focus, they darkened the overall exposure except for the player's face. Catch lights in the player's eyes were emphasized using a white brush and overlay blend mode. The speaker also addresses the limitations of generative fill for background creation and expresses optimism for future improvements in AI image generation.
📹 Conclusion and Invitation for Engagement
The speaker concludes by expressing excitement about the increasing options and possibilities for creative expression using AI tools. They clarify that they were not sponsored by any of the companies mentioned and made the video to share their recent learnings. The speaker encourages viewers to like, subscribe, and turn on notifications for more content. They also invite viewers to watch another video that YouTube recommends or explore some of their lighting tutorials.
Mindmap
Keywords
💡AI
💡Composite Images
💡Photoshop
💡Generative Fill
💡Firefly
💡Mid-Journey
💡Retouching
💡Lighting Setup
💡Action Shot
💡Neural Filters
💡Catch Lights
Highlights
The video shares the process of using AI to generate composite images of NFL football players.
Vato AI was used for skin retouching to remove blemishes and brighten dark circles under the eyes.
Firefly and MidJourney AI tools were utilized to generate backgrounds for the composite images.
Generative Fill in Photoshop Beta was employed to fix the backgrounds of the images.
Photoshop was essential for blending all elements of the composite images together.
The speaker's childhood interest in collecting trading cards influenced his career as a photographer.
The process involved photographing about 40 different athletes in a single day for trading cards.
Ellen Chrome mobile app was used to control lighting for different setups, allowing quick transitions.
Various lighting setups were used, including clamshell lighting and edge lighting techniques.
Action shots were created using a 70-centimeter OCTA box and strip softboxes for a dynamic effect.
The mobile app was programmed with different setups to facilitate rapid changes between shots.
Raw files were delivered to the client, who selected the images for their own purposes.
The speaker processed images in Capture One with a focus on high contrast for a punchy look.
Adobe Firefly was effective for creating heavily blurred cityscapes and landscapes.
MidJourney was better suited for generating detailed stadium and tunnel scenes.
The use of descriptive keywords in AI prompts was crucial for achieving desired background effects.
Neural filters in Photoshop were used to harmonize color and contrast between the foreground and background.
Techniques such as adding grain, adjusting exposure, and enhancing catch lights were used for final image polish.
The limitations of AI-generated image sizes were noted, with hopes for future improvements.
The video concludes by emphasizing the expanding creative possibilities with AI tools in photography.
