KDA Firestarter Photo Guide: Best Photos for Person or Cat

One subject, framed well

Every successful render depends on adhering to strict framing constraints. As detailed in this guide, the system requires exactly one photo featuring a solo subject. Submissions containing multiple individuals, background bystanders, or crowded group gatherings cause joint recognition conflicts, leading to distorted motion transfer or generation failure during the initial keypoint detection phase of processing across our automated pipeline.

Ensure your solo subject occupies the primary focal area of the frame. Human subjects should be captured in a standing or upright position with arms and torso clearly distinguishable from the surrounding backdrop. High-contrast separation between your clothing and background enables the neural network to extract clean silhouettes, preventing background elements from improperly warping into the synthesized dance routine during rapid movement sequences.

Person mode or cat mode

Users can select between two specialized generation variants: Person mode and Cat mode. The Person variant is calibrated for human anatomy, accurately tracking shoulders, elbows, hips, and knees throughout the intense routine.

In contrast, the Cat variant adapts the choreographic segment for feline subjects, translating Kai'Sa-style staging into four-legged motion dynamics. Both modes utilize the same twelve-second reference choreography, delivering an entertaining alternative for pet owners who want to see their cats perform stylized pop routine motions.

Photo selection directly determines the structural stability of the final clip. Uploads must feature only one photo in, depicting a single solo subject without background figures or heavy obstructions. For human subjects, full-body or three-quarter framing with even lighting yields the cleanest limb tracking.

When using Cat mode, ensure your pet is clearly visible against a contrasting background with paws and head unobstructed. Avoid blurry images, group portraits, or heavily filtered snapshots that obscure physical joint outlines.

Light and a clear face

Proper illumination provides the foundation for accurate depth estimation and keypoint localization. Following this guide, always utilize balanced, diffuse lighting that evenly exposes the subject without harsh shadows across the face or torso. Direct overhead glare or severe backlighting obscures facial features and complicates the model's ability to maintain subject identity during dynamic movement and rapid choreographic turns.

Avoid heavily filtered portraits, excessive beauty smoothing, or dramatic artistic vignettes. Natural daylight or neutral indoor room illumination produces consistent surface textures across the skin and clothing. Clear visibility of eyes, nose, and facial contours ensures that identity characteristics carry over smoothly into the 720p (or 1080p HD) MP4 output throughout the twelve-second performance without visual degradation or facial blurring.

Getting a good cat photo

The platform features a dedicated Cat mode designed specifically for feline anatomical structures. In this section of the guide, we recommend photographing your cat in a seated or standing posture where all four limbs and the tail are naturally positioned within the camera view. Avoid photos where the pet is curled tightly into a ball or tucked beneath heavy household blankets.

Contrast remains critical for feline subjects, particularly for dark-furred or patterned cats. Position your pet against a plain wall or solid-colored floor to prevent fur edges from blending into domestic furnishings. A clear, attentive gaze toward the lens assists the vision model in mapping facial landmarks onto the stylized twelve-second choreographic reference sequence with consistent physical tracking across the entire animation.

Only use photos you may use

Responsible content creation demands strict respect for individual privacy and image rights. According to the principles in this guide, users must only upload photos of themselves or individuals from whom they have obtained explicit prior consent. Utilizing photos of unconsenting acquaintances or third parties violates platform usage expectations and general digital ethics across online creator communities and social networks.

For pet owners utilizing Cat mode, ensure your animal companion was photographed comfortably without forced restraint or unnecessary physical stress. Ethical synthetic media generation relies on thoughtful user responsibility, ensuring that every synthesized animation reflects a respectful, consensual application of modern image-to-video machine learning capabilities across community platforms, personal media channels, and public social video distribution networks without compromising personal privacy or ethical standards.

Common mistakes

Certain image defects routinely compromise video quality or trigger automated pipeline rejections. Users following this guide should avoid low-resolution camera phone captures, compressed screenshots, motion-blurred snaps, or heavily cropped selfies. If a subject's hands or feet are cut off by the edge of the frame, the model must artificially extrapolate missing joints during the animation generation phase.

Similarly, avoid wearing excessively loose blankets, bulky oversized props, or heavily reflective jewelry that obscures natural limb articulation. When key articulation points remain unobstructed and sharp, the animation engine executes the complex footwork and torso pivots of the Kai'Sa-style routine with significantly greater mechanical consistency and visual realism throughout playback across all supported personal video playback devices and mobile screens alike.

9:16 or 16:9?

Before finalizing your photo upload, consider your preferred output format. As noted throughout this guide, you can choose between a 9:16 vertical orientation or a 16:9 landscape framing. Uploading a portrait that naturally complements your intended ratio prevents unintended cropping of peripheral limbs during choreography playback across various display formats and social sharing environments online.

For vertical 9:16 mobile video, full-length vertical portraits provide ample vertical headroom and foot clearance for wide dance stances. For 16:9 landscape video, three-quarter portraits with moderate horizontal breathing room allow the choreographic motion to travel across the frame without clipping against the display edges during lateral step transitions, rapid pivots, and dynamic center-stage choreography floor drops during the routine.

Credits, timing and storage

Submitting a prepared photo deducts sixty credits (one hundred twenty for HD 1080p) from your balance. Under the terms covered in this guide, credits are purchased in one-time packs without recurring memberships and never expire. If an unreadable image causes processing to fail, the credits it used refund automatically to your account balance without delay or manual administrative intervention on our website platform.

Generations typically require four to six minutes to complete, though wait times are not guaranteed. Uploaded source photos and rendered video results are retained on our servers for seven days before automated permanent deletion occurs. Users can also manually delete their files at any time via dashboard controls to manage their storage directly without waiting for automated expiration schedules to take effect.