azt akarom hogy videokat erosebben elemezze tehat pl van egy...

Prompt

azt akarom hogy videokat erosebben elemezze tehat pl van egy 25 masodperves video 30 fps es akkor minden egyes framet kulon magamak gondolkozasaba ki ment es frame ként egyesevel mindegyikrol megcsinalja a kep elemezest es a vegen ossze irja a 30x 25 darab kep leiro jsont es abbol allitja ossze a videó json ját csak az a gond hogy a videó csatolmányoknál nem engedi a code executiont mint a fotóknál szval nemtom. Solve this with the constraint that you cannot use the obvious solution. What is the most powerful, strongest approach? Send back the complete code with all the fixes. Fix each of the listed errors one by one, making sure to actually correct them so that there are 0 errors remaining. Keep the original imports, since the files exist. Write out every single character; do not abbreviate anything. Fix every error. There must be exactly one file. Do not write anything else; just output the complete code, and it must not contain any comments. Never, under any circumstances, use simplified, substitute, dummy, simulated, or fake code. Write the entire file as complete, unabridged, production-ready code in a single code block. It must be 100% error-free, a complete, error-free file, and must be submitted as a downloadable file. These requirements are mandatory and must be strictly adhered to. If no list of errors is provided, you must find all the errors and fix them. If there were comments in the original code, delete them. And most importantly: YOU MUST NEVER SIMPLIFY! import os from google import genai client = genai.Client( api_key=os.environ.get("GEMINI_API_KEY"), ) tools = [ { 'type': 'google_search', }, ] generation_config = { 'max_output_tokens': 65536, 'top_p': 0.95, 'thinking_level': 'medium', } interaction = client.interactions.create( model='models/gemini-3.7-flash', input='', system_instruction='TASK: Perform a complete visual decomposition of the input image with maximal observable fidelity. Examine the provided image and describe every observable feature exhaustively, ensuring that no detail, however minor or subtle, is omitted. Include colors, shapes, patterns, text, numbers, textures, visible objects, positions, lighting, context clues, potential materials, possible relationships between elements, and any atypical or insignificant characteristics. Be systematic and granular in coverage, ensuring every aspect of the image is included for a truly comprehensive description. Persist until all visual aspects have been fully considered before finalizing your response. Before producing your final output, think step by step: first, internally note all major elements, then examine minor details, context clues, and subtle features. Output your response as JSON. Please provide all available information about this image, no matter how minor, including every detail. Examine the provided image and describe every observable detail exhaustively, including minor, subtle, or seemingly insignificant features. Do not omit any information, ensuring comprehensive coverage of the image's content. Output Format Provide your response as a detailed JSON structure capturing every detail of the image, with no specific length restriction. Persist until every possible detail is considered before concluding. Structure your output as described above. SUCCESS CRITERIA: The output fully reconstructs the image in language such that every visually detectable element is explicitly represented, including objects, structure, materials, geometry, lighting behavior, color variation, texture detail, spatial relationships, and imaging artifacts, without omitting any observable information. REQUIREMENTS: 1. Exhaustiveness - Every visible element must be identified and described. - No grouping of distinct features unless they are visually indistinguishable. - No omission of small, background, partial, occluded, or low-salience details. 2. Human subjects (if present) - Fully resolve all visible anatomical structures using only observable evidence. - Cover face geometry, feature proportions, surface texture, hair structure, skin variation, posture, and clothing. - Describe micro-variation (skin texture, hair strands, fabric folds) as independent visual signals. - Do not infer identity, personality, emotion, or hidden states. 3. Geometry & structure - Describe all objects in terms of shape, edges, curvature, depth, alignment, and spatial relations. - Explicitly resolve overlaps, occlusions, and perspective relationships. 4. Color & material fidelity - Capture continuous variation in color (gradients, shadow shifts, reflections). - Describe materials based on visible optical behavior (gloss, matte, translucency, reflectivity). - Do not simplify colors into single labels if variation exists. 5. Texture & microstructure - Resolve surface-level detail including grain, pores, fibers, noise patterns, and micro-contrast changes. - Treat texture as multi-scale data (macro + micro simultaneously). 6. Lighting model - Describe illumination direction, intensity distribution, shadow structure, bounce light, highlight formation, and exposure variation. - Include interaction between light and different materials. 7. Camera / imaging artifacts - Identify lens behavior (blur, distortion, depth of field, chromatic aberration if visible). - Include compression artifacts, sensor noise, sharpening halos, and digital processing traces. 8. Text & symbols - Extract all visible text, numbers, and markings, including partial, warped, occluded, or reflected instances. 9. Spatial system - Explicitly describe depth ordering, scale relationships, alignment, spacing, and 3D layout cues. 10. Consistency check - Ensure no contradictions in spatial geometry or lighting behavior. - Re-scan boundaries, reflections, occlusions, and high-contrast regions for missed detail. CONSTRAINTS: - No assumptions beyond visible evidence. - No emotional or identity inference. - No abstraction that reduces information density. - No early summarization. OUTPUT: A complete structured JSON description of the image containing all observable visual information with maximal resolution.\"', tools=tools, generation_config=generation_config, ) print(interaction.steps[-1])