Image Physics Reality Check
Check images or prompts for physical realism
Image Physics Reality Check
Check images or prompts for physical realism
Showcase
Description
Review generated images or image-generation prompts for real-world physical accuracy. First, identify the image style and adapt the strictness of the review. Then, provide an itemized issue report with physical reasoning and severity levels. Finally, consolidate the findings into a revision prompt you can copy directly into an image-generation model, helping improve the physical realism of generated images from start to finish.
Related Skills
View allRealistic Skin Texture Prompts
Generate text-to-image prompts for selfies, close-ups, half-body portraits, and portrait photography, with a focus on reducing plastic-looking skin, an oily film, and AI-looking faces. Skin is guided toward more natural glow, texture, and subtle variation. Whether you need everyday portraits, magazine-style portraits, direct-flash snapshots, or realistic portraits with less of an AI look, prompts can be customized based on the subject, camera position, lighting, style, and preferred imperfections. Prompts are organized around the overall skin texture, highlight placement, differences across facial areas, textures such as pores and peach fuzz, and natural skin-tone variation, rather than simply stacking vague terms like “realistic skin” or “ultra-realistic.” You can specify image ratios such as 9:16, 3:4, and 1:1, and get composition and parameter suggestions for the image-generation model you use. When needed, you can also generate Chinese, English, or bilingual versions for GPT-Image, 即梦, 豆包, Midjourney, SDXL, or Flux. In addition to copy-ready positive prompts, the Skill explains how key descriptions affect skin texture and provides targeted troubleshooting and adjustment suggestions for over-smoothed skin, an overly oily face, a waxy appearance, or distorted texture. Generated prompts can also be saved by topic for easier lookup, fine-tuning, and consistent texture across a portrait series.
VideoRealistic Video Prompt Master
Used to create, rewrite, or review AI video generation prompts that produce visuals with a realistic live-action feel. Suitable for text-to-video and image-to-video prompt writing scenarios involving people, especially meeting user demands for realistic human feel, natural movements, reduced AI distortion, styles like Seedance / TikTok realistic style, believable character behavior, realistic skin texture, camera movement, lighting effects, natural interaction between people and environment, or writing negative prompts to avoid issues like plastic dolls, CG animation, and doll-like distorted images.
ImageViral Infographic Maker
Enter an object name and any optional requirements to receive a research-verified reference infographic about it. Whether it is machinery, an electronic product, a living organism, a building, a vehicle, or a natural object, the system organizes its English name, field, key parameters, structural components, geographic or industry distribution, development lineage, use cases, comparisons with similar objects, identification points, and maintenance recommendations, while recording traceable data sources. Information that cannot be confirmed is clearly marked “Unverified.” The finished design uses a low-saturation gray-blue industrial drafting style, with an exploded view of the main subject and structural cross-section diagrams. Key data is presented as parameter tables, ring charts, distribution maps, lineage diagrams, and comparison panels to help readers quickly understand the object’s appearance, construction, performance, and applications. For abstract or data-based subjects, suitable geometric diagrams or data visualizations are used instead. The final deliverable is a strictly 3:4 infographic poster suitable for high-resolution viewing and print, along with an editable HTML source file. The layout includes title identification, main subject illustrations, structured information modules, and source notes, with a focus on complete content, clear text, and verifiable data.
Information
- Version
- v3
- Last updated
- Runtime credits
- Usage-based
- Models
- Auto
