Selfie Guides
Tableau Selfies Guide
what’s different in tableau tableau is highly flexible and follows natural language closely it supports both very short and long prompts (up to 2000 characters), but length is not rewarded by itself the goal is control, not filling the character limit clear, functional descriptions work better than poetic or decorative language, and you do not need to over explain concepts the system already understands pose reference you can upload (or paste in) an image to be the pose reference for a given prompt pose reference tries to match the output image as close to the given pose as much as possible with pose reference, even the body shape of the main subject(s) will be taken into consideration in the output to match close to the reference image pose references are the most powerful way to control the composition of the output technical note if the pose reference image for group selfies is too complex, you are less likely to get the desired result prompting for style a wide range of art styles are supported while maintaining photoreal avatar consistency different photographic looks, lighting styles, moods, and aesthetics can be applied without breaking facial likeness, including in group selfies unlike earlier versions, tableau largely resolves facial consistency issues when styles change, allowing experimentation without sacrificing realism prompt variety is handled natively , so the enhance variety prompt is no longer needed how to write a strong avatar description (most important step) the avatar image helps map facial features it matters more for photoreal avatars, especially when paired with a detailed description for anime avatars, the image is less important, and the description is the primary factor in achieving an accurate character avatar image should be clean slightly angled face clearly visible no tattoos no face obstruction think of the avatar as a neutral reference, not a styled photo avatar description avatar descriptions can be very detailed do not describe the facial structure for photoreal in this space, use face detail (okay for anime) use this structure as your baseline template age gender ethnicity/race skin (tone + texture details) hair (length + style + color) eyes (use “natural eyes”) face (shape, nose, lips, brows) build (body type) details (tattoos, scars, jewelry, piercings) style detailed textures, soft cinematic lighting example a 24 year old woman with pale skin and a soft natural flush, she has long wavy dark brown hair threaded with caramel highlights, gentle features with a straight nose and full lips, and a curvy, toned build she wears a small silver nose stud, shown in soft cinematic lighting face details (separate from ad) eye color, freckles, scars, skin texture, pores, wrinkles belong in the face detail prompt, not the avatar description treat face details as adjustable render parameters, not identity anchors natural green eyes and slight freckles across her nose face detail slider (important) 0% face detail very strong adherence to the avatar picture, only minimal adherence to face detail prompt faces look smoother and slightly plasticky / less realistic higher face detail values more pores, micro details, and realistic facial shadows higher risk of drifting away from the avatar image why this matters tableau will follow this description more strictly than any previous version if something is missing or vague here, autoselfies will drift keep everything literal no metaphorical marks (“star marking” will create a literal star) use descriptors like vascularity for defined muscles add lighting cues inside avatar description if you want consistency how to prompt since weights, negative prompts, wand, and pose refs are gone, tableau requires straightforward description core rule start with what matters most → move to scene → finish with style/lighting structure identity anchor (optional but useful) action / pose setting clothing camera / lighting / mood example \[name] stands outdoors, facing the camera with a relaxed and natural expression they are wearing \[clothing description] the environment around them shows \[background elements] lighting is \[type] and the image has \[texture/detail cues] keep related details close together if you mention a campfire and smoke, phrase it like smoke rising from the campfire instead of separating the details across the prompt avoid these (deprecated or unsupported) any (weights 1 2) syntax negative prompts “///” formats generating autoselfies before using autoselfies, follow this checklist check your avatar description first it now determines face consistency skin tone hair shape overall realism stylistic cohesion practical prompting tips 1\ describe the whole scene to avoid zoomed in closeups if you want the model to “zoom out,” describe legs, shoes, and the background 2\ use natural language intensity examples “ extremely detailed braid” “ very soft natural lighting” “ subtle freckles across the nose” 3\ use strong lighting cues phrases like ' detailed textures ' and ' soft cinematic lighting ' are almost essential for avoiding flat or lifeless renders 4\ group related ideas together example a bonfire lighting the campsite, soft smoke drifting upward works far better than scattering those details into separate fragments 5\ for non human characters species must be declared first details must be literal avoid stylized metaphors unless you want artificial shapes example selfie prompt (starter template) you can adapt this as your universal starter a natural candid selfie of \[avatar name] standing outdoors they face the camera with a relaxed, genuine expression they are wearing \[clothing description], and the background shows \[environment] lighting is soft and cinematic, highlighting natural textures in the skin and hair the image feels like a real photo with detailed textures and clean realism example prompt for more variation \[avatar name] takes a selfie at golden hour, the warm sunlight creating soft highlights on their face they wear \[clothing], and behind them, the sky glows with fading orange light the wind gently moves through their hair, and the camera captures a wide view of the scene cinematic lighting, clear background details, and natural depth give the photo a grounded realism when something goes wrong use this quick diagnostic problem quick fix identity drift strengthen the avatar description with necessary details to improve accuracy, including skin texture and tone cues use the face detail section for all elements related to facial features (photoreal) too zoomed in add full body clothing details add surrounding scenery add environment objects lighting looks flat add explicit lighting details include phrases like “detailed textures” or “soft cinematic lighting” in the avatar description