.gif)
Scenario streamlines numerous creative pipelines in game design, including the rapid development and generation of characters for portrait-style talking head videos. This efficiency allows you to allocate more time to story development and quickly edit your animated portraits to incorporate new themes and narratives.
These portrait stills, which represent communication between player characters and NPCs are pivotal. They often adhere to distinct design rules separate from the main game and provide an excellent opportunity to incorporate storytelling elements within your game.
.jpg)
This workflow is accessible to designers of all skill levels and involves simple style training. We recommend following the workflow one step at a time, as outlined below:
1. Setting Up a Model
2. Generating Base Portraits
3. Creating New Expressions in Canvas
4. Finalizing with Sketch & Erase
For this example workflow, we will be using a pre-created platform model. Should you prefer to explore your own style or character, some steps in the workflow may need to be modified. It is recommended to follow these steps first with the Fantasy Girl Portraits platform model, if you are new to Scenario.
.jpg)
You can find the Fantasy Girl Portraits model by navigating to the Models page and looking at the Platform Models tab, or by following this link directly to the model. Once on the model page, simply click Generate with This Model to get started.
The next step is straightforward - create your base character portrait. This starts by crafting a prompt that describes the type of character you prefer. The model we have chosen is suitable for visual novel games, point and clicks, or 2D talking head NPCs. Some examples of good prompts are:
‘Auburn curls, light blue eyes, tanned skin, beast master, vest with multiple animal motifs ’
‘a pirate queen with olive skin and green eyes, curly hair’
‘A willowy elf female with alabaster skin, crystal blue eyes, and platinum hair reads an ancient tome in a mage’s cloak’
%20copy.jpg)
You can also incorporate words that modify or emphasize specific stylistic elements in the model.
Once you have generated an image you are happy with, go on to Step 3.
Now it is time to generate the various facial expressions for your animation. The most straightforward way to do this on Scenario is by using our Canvas feature. Simply select your favorite base image from Step 2, click the … button on the top bar, and select Edit in Canvas. This will transfer the image, prompt and model information to Canvas.
.jpg)
Generating Facial Expressions
Begin by clicking the paintbrush shaped tool from the vertical toolbar next to the inference settings. It is the 4th icon from the top. This will allow you to paint a mask on your character portrait, blocking out areas that can be regenerated or ‘inpainted’.
First, mask the mouth and a few small areas around the face. Try to keep your masked areas as small as possible to maintain the best consistency. You want to make sure to target:
Next, adjust your Prompt and add a description of the new expression at the beginning. For example, you could use "open mouth," "laughing," or "angry" to guide the AI's output. After modifying the prompt, click Reference Image and adjust the influence to a value between 8 and 18. The lower you move the slider, the more the image will deviate from it’s reference.
Click Generate to create the new image based on your adjustments. Make any necessary tweaks and repeat the process as needed until you achieve the desired result. We recommend copying and selecting your base image, selecting the new expression image and clicking Merge Layers to ease the following steps. This will end up creating a new complete layer for each expression.
Creating Closed Eyes
To modify the eyes, select the Sketch tool from the vertical toolbar, followed by the Color Selector. Use the Eye Dropper tool to pick a color that matches the desired eyelid color, then carefully paint over the eyes.
Next, switch to a color that matches the eyelash color you want, reduce the size of the paintbrush, and create a rough sketch of the eyelashes.
Once you've made these adjustments, switch back to the Mask tool. Mask out the roughly adjusted eyes, then add "Closed eyes" to the beginning of your prompt. Increase the Reference Image influence to approximately 25, then click Generate. Adjust the influence as needed to achieve the desired result.
In this section, we will mix and match facial features by removing unwanted areas and merging images. Begin by using one of your newly merged images and making a copy of it that you can edit.
Ideally, you should have two layers: the bottom layer should have the closed eye image and the top layer should have a copy of the new expression. We recommend that both layers you edit in this step are copies of existing layers.
%20copy.jpg)
Position the new expression above the closed eye layer. Make sure the expression layer is selected. Using the Erase tool, which can be found n the vertical toolbar, erase the eyes. This will revel the new closed eyes underneath.
If you're meticulous, you may not need to mask anything out. However, if you find that some edges don't look quite right, switch again to the Mask tool and mask out the problematic edges. Set your Influence a bit higher, between 30 and 50, to avoid adding strange details.
Once you've merged your images, export them all at once by clicking Download and then Save All Layers. Ensure that all the layers you want to download are visible before proceeding.
%20copy.jpg)
Canvas also allows you to easily remove backgrounds as needed. However, it's recommended to do this after completing all image edits, as removing the background prematurely may complicate further masking and generation. To remove the background directly in Canvas, click the layer you want to target and select Remove BG.
Now you have all the images you need to pull together as animation stills in your preferred animation program. Depending on time and resources, this workflow can be leveraged for more complex animation details, such as movement and changes in different piece of clothing or background scenes.
Regardless of studio size, time is always a critical factor in game development. This workflow can help you enrich your game story and world, as well as create opportunities for your LiveOps and Marketing teams to develop ads or introduce new characters without disrupting the consistency of your game world's style.
.svg)
.gif)
.png)
