This Background is FAKE: Transforming Backgrounds in DaVinci Resolve & Photoshop with AI | Tutorial

Josh Chappell
12 Oct 202307:33

TLDRIn this tutorial, the host demonstrates how to transform a video background using DaVinci Resolve and Adobe Photoshop's generative AI. Starting with a color graded clip, the process involves using the magic mask tool to select the subject, tracking the mask, and replacing the background with a new image. The host then uses Photoshop's generative fill to create a new background that matches the original footage's lighting and focal length. Finally, adjustments are made in DaVinci Resolve to blend the new background seamlessly, resulting in a convincing final image.

Takeaways

  • ๐ŸŽฅ The tutorial demonstrates how to transform a video background using DaVinci Resolve and Adobe Photoshop's generative AI.
  • ๐ŸŒˆ A color graded clip is used, filmed with the Alexa Phantom L on a Sony a74 with SLOG3.
  • ๐Ÿ“‚ Clips are imported into DaVinci Resolve and a small section is selected for the tutorial.
  • ๐Ÿ–ผ๏ธ The magic mask tool in DaVinci Resolve is used to create a detailed outline around the subject.
  • ๐Ÿ”„ Mask tracking is applied to the clip to capture movement throughout the footage.
  • ๐Ÿ“บ An Alpha output is added to the masked clip to prepare it for background replacement.
  • ๐Ÿ–Œ๏ธ The tutorialist duplicates the clip and resets the node grade to start fresh for the new background.
  • ๐Ÿ–ผ๏ธ A frame is selected and exported as a TIFF file to be used in Photoshop for background generation.
  • ๐Ÿ’ก Photoshop's 'Select Subject' is used to create a rough selection of the subject for background replacement.
  • ๐ŸŒŸ Generative Fill in Photoshop is utilized with a detailed prompt to create a matching background.
  • ๐Ÿ”ง Final adjustments are made in DaVinci Resolve to blend the generative AI background with the original footage.

Q & A

  • What is the main topic of the tutorial?

    -The main topic of the tutorial is transforming backgrounds in DaVinci Resolve and Photoshop using AI technology.

  • Which software is used to replace the background in the footage?

    -DaVinci Resolve and Adobe Photoshop are used to replace the background in the footage.

  • What type of AI is mentioned in the script for background replacement?

    -The script mentions using generative AI for background replacement.

  • What camera and settings were used to film the original footage?

    -The footage was filmed on a Sony a74 with SLOG3 settings.

  • How does the tutorial suggest improving the computer's understanding of the subject for masking?

    -The tutorial suggests making detailed outlines around the subject, such as the hands and hair, to give the computer more information.

  • What is the purpose of tracking the mask information in the tutorial?

    -Tracking the mask information allows the mask to follow the subject throughout the clip, ensuring the background replacement is consistent.

  • Why is it important for the AI-generated background to match the color and lighting of the original footage?

    -Matching the color and lighting ensures that the AI-generated background looks natural and blends well with the subject.

  • How does the tutorial suggest ensuring the focal length of the AI-generated background matches the original footage?

    -The tutorial suggests shrinking the subject in Photoshop to 'trick' the AI into generating a background with the correct focal length.

  • What is the role of the 'generative fill' feature in Photoshop during the background replacement process?

    -The 'generative fill' feature in Photoshop is used to create a new background based on a given prompt, which helps in matching the lighting and color of the original footage.

  • Why does the tutorial mention making the subject small in Photoshop before using generative fill?

    -Making the subject small in Photoshop before using generative fill helps the AI to generate a background with the correct lighting and focal length as if the subject were full size.

  • What is the final step in DaVinci Resolve to make the AI-generated background blend with the footage?

    -The final step is to apply lens blur to the AI-generated background to make it blend more naturally with the footage.

Outlines

00:00

๐ŸŽฅ Green Screen Keying and Background Replacement

This paragraph describes a video editing process using Da Vinci Resolve and Adobe Photoshop's generative AI. The speaker starts by explaining how they applied a LUT to color grade their footage, shot on a Sony a74 with SLOG3. They then import the clips into Da Vinci Resolve and use the magic mask tool to create a detailed outline around their hands and other parts of their body to provide the computer with more information for accurate keying. The speaker emphasizes the importance of tracking the mask information throughout the clip and setting an appropriate in to out ratio and clean black level. The goal is to create a convincing masked-out image that can be used to replace the background with a new one, which doesn't have to be perfect to fool the eye. The process involves duplicating the clip, resetting the node grade, and exporting a still image to be used in Photoshop for generating a new background that matches the lighting and focal length of the original footage.

05:05

๐Ÿ–ผ๏ธ Creating a New Background with Generative AI in Photoshop

In this paragraph, the speaker details the process of creating a new background for their footage using Adobe Photoshop's generative AI feature. They start by selecting a frame where their hands are not obstructing the view and use the 'select subject' tool to isolate themselves from the original background. The speaker then shrinks their image to a very small size to trick the AI into generating a background that matches the lighting and focal length as if the person were full size. They use the generative fill feature with a prompt that includes details about the desired background, such as a large living room space with white walls and natural light coming from the left side. After generating the background, the speaker is satisfied with the result, noting the gobo lighting and balanced texture. They mention that the AI has managed to create a background with a deeper focal length that matches the original footage. The speaker then brings the generated background back into Da Vinci Resolve and applies lens blur to the AI image to make it blend more naturally with the keyed footage. The final result is a convincing replacement of the original background with a new one generated by AI.

Mindmap

Keywords

๐Ÿ’กDaVinci Resolve

DaVinci Resolve is a professional video editing software that offers color correction, visual effects, and audio post-production features. In the context of the video, it is used to replace the background of footage using its advanced color grading and masking tools, which are essential for creating a realistic composite.

๐Ÿ’กAdobe Photoshop

Adobe Photoshop is a raster graphics editor that is widely used for photo manipulation, editing, and creation. In the video, it is utilized in conjunction with DaVinci Resolve to create a new background for the footage using its generative AI capabilities, which allow for the creation of custom environments.

๐Ÿ’กGenerative AI

Generative AI refers to artificial intelligence algorithms that can create new content, such as images, based on existing data. In this video, generative AI within Photoshop is used to generate a new background that matches the lighting and perspective of the original footage, creating a seamless composite.

๐Ÿ’กMagic Mask Tool

The Magic Mask Tool in DaVinci Resolve is a feature that simplifies the process of creating masks for complex subjects. In the video, it is used to select and isolate the person from the original background, which is a crucial step in replacing the background with a new one.

๐Ÿ’กColor Grading

Color grading is the process of adjusting and enhancing the color of a video to achieve a specific look or mood. The video mentions that a color graded clip is used, which means the original footage has already undergone this process, and the new background must match this grading to maintain visual consistency.

๐Ÿ’กAlpha Output

Alpha Output in DaVinci Resolve refers to the process of creating an alpha channel for a clip, which defines the transparency of the image. This is essential when replacing backgrounds, as it allows the software to distinguish between the subject and the background.

๐Ÿ’กClean Black

Clean Black is a setting in DaVinci Resolve used to remove any remaining background noise or color after a keying operation. In the video, it is used to refine the mask and ensure that the subject is cleanly separated from the original background.

๐Ÿ’กTIFF File

A TIFF (Tagged Image File Format) file is a type of image file format that supports high-color depth and is often used in professional image editing. In the video, a TIFF file is exported from DaVinci Resolve to be used as a reference for creating the new background in Photoshop.

๐Ÿ’กFocal Length

Focal length refers to the distance between the camera lens and the image sensor when the subject is in focus. It affects the field of view and perspective of the image. The video emphasizes matching the focal length of the new background to the original footage to maintain a natural look.

๐Ÿ’กGenerative Fill

Generative Fill in Photoshop is an AI-powered feature that can create synthetic images based on textual descriptions. In the video, it is used to generate a new background that matches the lighting and perspective of the original footage, creating a realistic composite.

๐Ÿ’กLens Blur

Lens Blur is an effect in DaVinci Resolve that simulates the natural blur caused by a camera lens. It is used in the video to adjust the depth of field of the new background, helping to integrate it with the original footage and create a more convincing composite.

Highlights

The tutorial demonstrates how to transform backgrounds in DaVinci Resolve and Photoshop using AI.

The footage is initially against a closet wall and will be replaced with a new background.

A color graded clip is used, applying an LUT to achieve the desired look.

The footage was filmed on a Sony a74 camera with SLOG3 settings.

The clips are imported into DaVinci Resolve for editing.

A small clip is selected for the tutorial to keep it concise.

The magic mask tool in DaVinci Resolve is used to select parts of the footage.

Detailing the mask around the subject provides the computer with more information for processing.

Mask overlay is toggled to track the selected information throughout the clip.

An Alpha output is added to the clip for masking purposes.

The clean black setting is adjusted to improve the mask quality.

The background replacement does not need to be perfect to be convincing.

The tutorial then moves to duplicating the clip and resetting the node grade.

A frame without much hand movement is selected for exporting as a TIFF file.

Adobe Photoshop is used for AI-based background generation.

The 'Select Subject' tool in Photoshop helps in creating aๅˆๆญฅ selection of the subject.

Lighting and focal length matching are crucial for a realistic background่žๅˆ.

A trick is used to shrink the subject in Photoshop to match the background's perspective.

Generative Fill in Photoshop is utilized to create a new background based on a prompt.

The background prompt includes details like 'large living room space with white walls' and lighting direction.

Tweaking the generative fill settings can improve the background's realism.

Leading lines and lighting can be included in the prompt to enhance the composition.

The AI generative fill is fooled by making the subject small, ensuring proper lighting and focal length.

The final background is imported back into DaVinci Resolve for blending.

Lens blur is applied in DaVinci Resolve to make the generative AI image blend better.

The final image's convincingness is left for viewers to judge in the comments.