Image 2.5 Masterclass: From Personal Fun to E-commerce Success (Full Prompt Guide)

1Please transform the image I uploaded into a standalone sticker suitable for making animated emojis.
2
3Automatically identify the main subject in the image:
4For people, retain recognizable facial features, hairstyle, clothing, and expression;
5For animals, retain breed appearance, fur color, patterns, and expression;
6For objects, retain unique outlines, structures, and main color schemes;
7If the image is primarily a scene, design the entire scene into a simple outline scene sticker.
8
9Adopt a slightly 3D cartoon sticker style with clear outlines, full shapes, and a touch of humor. Do not unify different subjects into the same face, and do not force limbs onto objects that originally don't have them.
10
11There should be a clean white die-cut border around the sticker. The subject should be centered with enough space around it, not cropped.
12
13Generate only one sticker, not a 3x3 grid, no additional text, and do not show the sticker applied to other objects.
14
151:1 square canvas, output a PNG with a real Alpha transparent background; do not draw the transparent checkerboard into the image.

1Please read the previous sticker image and use Python to create a truly playable looping GIF.
2
3First, check if the image has a real transparent background; if it has an opaque background, separate the subject from the background first, and do not let the background move with the subject.
4
5On a fixed pure white canvas, have the subject complete a light bouncing motion:
6Stay still for a moment, press down slightly to gather strength, jump up, flatten slightly upon landing, and then return to the original state.
7
8Use only the same sticker material, creating animation through position, slight rotation, and small scaling. Do not redraw the character frame by frame, and do not change facial features, patterns, colors, or clothing.
9
10The range of motion should be moderate, the transition smooth, and the subject should always remain completely within the canvas with a natural loop. The background should remain static without flickering or ghosting.
11
12Output a 512×512 pixel, approximately 2-second, infinitely looping GIF file. Actually check that the file contains multiple frames before providing the final product; do not just return code or a static preview.
13
14If the current environment cannot run the code or export the file, please state so clearly.

1Create a Meme sticker pack based on the images attached, remixing with 😎😛💕🚀🥳. Use exaggerated internet-reaction expressions, including crying, confusion, shock, smugness, side-eye, and deadpan disbelief, with awkward poses, low-fi cutout textures, and absurd humor. 
2Create a single square(1:1) transparent sticker sheet with nine distinct stickers arranged in a 3×3 grid, each showing a different expression, pose, or reaction. Separate the stickers with wide, fully transparent gaps. No background, shadows, or overlapping elements.

1Please transform the main person, animal, or cartoon character in my uploaded image into a set of exclusive chat emojis.
2
3First, extract the most recognizable features of this subject, then unify them into a simple, vivid, and slightly exaggerated hand-drawn sticker style. Retain facial relationships, hairstyle or fur, signature patterns, clothing, and main colors; do not turn it into a generic character.
4
5Generate a neatly arranged 3×3 grid image. The character in the nine grids must be the same, with consistent appearance and art style.
6
7From left to right, top to bottom, show in order:
8Laughing happily, full of confusion, completely speechless;
9Suddenly shocked, aggrieved and wanting to cry, secretly proud;
10Exhausted, pouting in anger, expressing love.
11
12Each grid must have a distinctly different expression and posture that can be understood even when shrunk. The movements should suit the subject's own body structure; do not force realistic human bodies onto animals.
13
14Leave white space around each sticker, they should not touch each other, no divider lines, and no text. Pure white background, 1:1 square.

1Please use the image generation tool, using the single image I uploaded as a character reference, to generate a six-frame continuous action sequence image for making a looping GIF.
2
3This step requires actually generating new action frames. Do not use Python to translate, rotate, scale, or locally stretch the original image instead of drawing.
4
5【Character and Action】
6
7Identify the main person, animal, or cartoon character in the image, retaining its recognizable facial features, hairstyle or fur, patterns, clothing, accessories, and main colors.
8
9If the original image is already a sticker or illustration, follow the original style.
10If the original image is a photo, unify it into a simple, vivid hand-drawn sticker style.
11
12This time, only create one expression/action:
13【Fill in here, e.g., Laughing happily; leave blank to automatically select a suitable action based on the uploaded image】
14
15Focus only on this one action; do not draw six unrelated expressions.
16
17【Six-Frame Continuous Performance】
18
19From left to right, top to bottom, present six continuous stages of the same action:
20
21Frame 1: Starting posture, action not yet fully unfolded.
22Frame 2: Beginning to move, face and related limbs start to change.
23Frame 3: Action unfolds, expression significantly strengthened.
24Frame 4: Peak of the action, emotion most clear.
25Frame 5: Action subsides, gradually recovering.
26Frame 6: Close to the starting posture, preparing to loop back to Frame 1.
27
28Please specifically design the mouth shape, cheeks, eyes, and limb postures for each frame based on the selected action.
29
30At least one facial feature must show clear, continuous change, with body movements as support. For example, when laughing, the mouth gradually opens and cheeks lift; when surprised, the mouth opens wide and the chin drops.
31
32Allow redrawing the same character's facial features and limb postures at different points in time. Keep the character's identity and styling consistent; do not lock the eyes, mouth, and posture completely.
33
34If the character wears sunglasses, keep them; express emotions through the mouth, cheeks, chin, and limbs, do not hallucinate eyes through the glasses.
35
36Animal movements must conform to their own body structure.
37
38【Inter-frame Consistency】
39
40Fix the camera, perspective, lighting, drawing style, and the character's base size.
41
42Use the same composition coordinates and body base position in each grid. Actions can change local outlines, but do not zoom the character in or out frame by frame just to fill the grid.
43
44Clothing, accessories, number of limbs, and fur patterns must not change without reason.
45Existing props and decorations can change reasonably with the action; do not randomly add or remove them.
46
47【Sequence Image Layout】
48
49The entire image is 3 columns × 2 rows, total aspect ratio 3:2, each grid is an equal-sized square.
50
51Use a regular invisible grid, no divider lines, outer frames, text, or frame numbers.
52
53Place only one complete character in each grid, including its props and decorations. Leave about 15% safety margin; no part should cross into adjacent grids.
54
55Use real Alpha transparency for the background; do not draw a white base or transparent checkerboard. Retain the sticker's white border and white details inside the subject.
56
57【Delivery】
58
59Directly generate this six-frame action sequence image; do not make the GIF yet.

1Please use Python to read the previously generated six-frame action sequence image file and split and synthesize it into a truly playable looping GIF.
2
3This round only processes the six action frames already generated; do not redraw the character or use the original static sticker to replace these frames.
4
5【Reading and Splitting】
6
7Confirm that a readable six-frame original image exists, then split the six frames in the order of 3 columns × 2 rows, from left to right, top to bottom.
8
9Check if each frame is complete and if it includes content from adjacent grids.
10
11Retain the character, limbs, tail, accessories, props, sticker white border, and all decorations belonging to that frame.
12
13Check if it has a real Alpha transparent background.
14
15If the original image has a solid opaque background, only separate the outer background, retaining the whites of the eyes, teeth, highlights, white clothes, and sticker white border. Do not directly delete all white pixels, and do not treat the checkerboard as a transparent background.
16
17【Unified Canvas and Alignment】
18
19The output canvas for all six frames is 512×512.
20
21Prioritize retaining the unified coordinate relationship of each grid in the original sequence image.
22
23When cropping or resizing is needed, use the combined boundary of the complete content of all six frames to determine a common cropping range, and use the same scaling ratio and position offset for all six frames.
24
25Prohibit cropping tightly to the outline frame by frame and then scaling each to maximum size.
26Prohibit re-centering frame by frame, which destroys the original movement displacement.
27
28Only correct obvious layout deviations, retaining intentionally designed head and limb movements. Ensure all frames have sufficient safety margins.
29
30【GIF Synthesis】
31
32Loop playback directly in the order 1→2→3→4→5→6→1.
33
34Retain the six actually generated content frames; do not add overall sticker drifting, rotation, or scaling effects.
35
36Default frame durations are:
37350, 200, 250, 450, 400, 350 milliseconds.
38Total duration 2 seconds.
39
40Pauses can be fine-tuned based on the actual action, but the final actual duration must be reported.
41
42Do not repeatedly copy a large number of static frames to reach 20 frames/sec.
43Do not use dissolves, transparency cross-fades, or motion blur to fake movement.
44
45Use a unified palette established based on all six frames, reserve a dedicated transparent color index, and correctly set the transparent color, background index, and frame disposal method. Use infinite looping to avoid ghosting, trails, and background flickering.
46
47【Actual Inspection】
48
49Re-open the GIF after exporting and check:
50Actual frame count, duration of each frame, total duration, size, infinite loop setting, transparent areas, and file size.
51
52Check frame by frame if the character goes out of bounds, if decorations are lost, and if there is a noticeable jump when the last frame loops back to the first.
53
54Observe the face and limbs simultaneously to confirm the playback content comes from six different action states. Do not claim expression changes are qualified based solely on pixel change rates.
55
56If the six frames themselves lack action changes or character consistency is obviously wrong, please point out the specific problem; do not cover it up with overall movement.
57
58【Delivery】
59
60Provide a link to a standalone GIF file.
61
62Also provide a ZIP containing the six split PNGs and the GIF.
63
64Briefly report the actual frame count, size, total duration, and file size read after re-opening the file.
65
66Run and export the final product directly; do not just return code, commands, or static previews. If you encounter situations where materials cannot be read or exported, please state so truthfully; do not provide fictional links.

1Please use the main product in the image I uploaded as the sole product reference to generate a high-quality product base image for subsequent dynamic advertisement production. Please strictly retain the original product's outline, proportions, structure, color, material, Logo, brand name, label text, packaging information, and all recognizable details. Do not redesign the product, and do not add text, selling points, prices, parameters, decorative elements, or accessories that do not exist in the original image.
2
3The image should adopt a high-end commercial product photography style, with the product facing the camera, centered, and the lens fixed as if on a tripod. Maintain natural and realistic perspective and product proportions, leaving plenty of safety space around the product. Use a simple, neutral, low-interference studio background; you can automatically choose off-white, light gray, warm gray, or other suitable neutral colors based on the product's own color, but do not use complex scenes, props, or obvious textures.
4
5The lighting should use soft, restrained, and uniform basic studio lighting. Do not add obvious scanning light, strong strip highlights, dramatic shadows, or local overexposure at this time. The overall exposure of the product should remain balanced, keeping the real texture of materials like glass, metal, plastic, fabric, leather, and liquid. Metal can have natural reflections, matte materials remain soft, and transparent or translucent materials retain a realistic sense of light transmission; do not change the product's original material to enhance the sense of high-end.
6
7This image will serve as the completely fixed base frame in the subsequent animation, so please prioritize ensuring the product structure, Logo, labels, and text are accurate and stable. The final image should be clean, restrained, and high-end, like a commercial product photography raw shot with lighting completed, where only moving lights need to be added to this fixed image later.

1Please use Python to turn the previous continuous keyframe collage into a truly playable looping GIF. The previous collage is fixed at 5 columns × 2 rows, totaling 10 frames; please crop strictly according to this grid, do not automatically change to other arrangements. The first row from left to right is frames 1, 2, 3, 4, 5, and the second row from left to right is frames 6, 7, 8, 9, 10.
2
3After cropping, keep all frames at a completely consistent canvas size and retain the original content proportions. Prohibit non-proportional stretching of the product; do not widen, flatten, or recompose the product. If the original single-frame sizes are different, you can only scale the entire single frame proportionally and then center it with padding to a unified size; do not move the product individually, and do not modify the product coordinates in each frame to "correct" the position.
4
5The GIF uses a fixed back-and-forth playback order: 1→2→3→4→5→6→7→8→9→10→9→8→7→6→5→4→3→2, then back to frame 1. Do not repeat frame 1 and frame 10 at the ends; this avoids the light suddenly jumping back to the far left after reaching the far right. Each frame is fixed at about 120 milliseconds, set to infinite loop, and the overall rhythm remains slow, smooth, and high-end.
6
7Export the GIF using a single unified 256-color palette for all frames to minimize color flickering caused by individual quantization of each frame. Do not use cross-dissolves, transparency overlays, optical flow interpolation, AI frame interpolation, inter-frame blending, or morphing to cover up problems from the generation stage, and do not add displacement, rotation, scaling, or deformation to the product at the GIF stage. The GIF stage is only responsible for cropping frames, unifying sizes, unifying the palette, arranging the playback order, and exporting the file.
8
9Upon completion, please actually read the generated GIF file to check, confirming the GIF contains 18 frames, each about 120 milliseconds, and has been set to infinite loop. Also check if the product's outer outline in the 10 original keyframes has position drift, if the Logo, brand name, and label text jump, if the product size changes, if there are obvious brightness sudden changes or color flickering between adjacent frames, and if there are obvious breakpoints in the entire back-and-forth loop.
10
11Please use image analysis to estimate the maximum inter-frame position offset of the product edges and the Logo/label area, but do not automatically correct these offsets. If the source keyframes themselves have jumping, please state truthfully that this is an inter-frame consistency issue from the generation stage; do not hide it through post-processing.
12
13You must ultimately output a truly playable GIF file and tell me the number of original keyframes, the actual number of GIF frames, single frame size, duration per frame, loop method, actual playback order, estimated maximum offset of product edges, estimated maximum offset of Logo/labels, and whether there is obvious flickering or loop breakpoints.

1Please edit Image 1, replacing the main product in it with the product in Image 2.
2
3The uses of the two reference images must be strictly separated:
4Image 1 is only responsible for the advertisement composition, background, lighting direction, decorative elements, and text layout outside the product.
5Image 2 is only responsible for the new product's real appearance, structure, material, color, Logo, and packaging information.
6
7Remove the old product from Image 1 and naturally place the product from Image 2 into the original product's area.
8
9Retain the real aspect ratio of the Image 2 product; do not flatten, stretch, or modify the structure to fit the original outline. You can adjust the overall size proportionally to make the new product fit the original layout.
10
11Only modify the area occupied by the new product and the necessary contact shadows, occlusions, and local reflections generated by the replacement. Do not redo the entire poster.
12
13The headlines, text content, fonts, font sizes, positions, background textures, decorative elements, and canvas size outside the product remain unchanged.
14
15Prohibit mixed designs of the two products. Do not retain the bottle cap, handle, label, or other parts of the old product.
16
17Do not add accessories not in Image 2, do not guess unclear packaging text, and do not change the product color to match the background.
18
19Output only one complete replaced poster.

1Following the poster template just confirmed, only replace the main product with this new one I uploaded.
2
3The newly uploaded photo is the sole basis for this round's product appearance. Do not follow the structure, labels, or colors of the previous one.
4
5The rest of the layout, copy, background, and decorations remain unchanged, only updating necessary shadows and reflections.

1Please put the top from Image 2 onto the person in Image 1 to generate a realistic try-on effect image.
2
3Image 1 is only responsible for the person's identity, body proportions, posture, camera angle, and background.
4Image 2 is only responsible for the top's style, color, fabric, cut, pattern, and details.
5
6Retain the person's face, hairstyle, skin tone, body type, and original posture, only replacing the top. Do not change shoulder width, waistline, or body proportions to make the clothes look better.
7
8Accurately retain the top's neckline, sleeve type, placket, pockets, buttons, and other visible structures.
9
10The original prints, embroidery, Logo, and text should be located in the corresponding positions on the clothes. They can naturally curve and wrinkle with the fabric, but cannot be rewritten, mirrored, added, deleted, or moved to other positions.
11
12The fabric should drape reasonably according to the shoulders, arms, and torso. Retain the original fabric's thickness and surface texture; do not turn knitwear into smooth plastic, and do not turn stiff fabric into thin, clingy cloth.
13
14Use a front display; do not actively show back designs not provided in the reference image.
15
16Retain the original image's light and environment, letting the replaced clothes blend in naturally. Output only one try-on image, no text or before-and-after comparison.

1Please place the transparent or semi-transparent product I uploaded into a new commercial photography scene.
2
3Retain the original product's outline, proportions, thickness features, bottle mouth or opening structure, label position, and recognizable packaging text. If there is liquid inside, retain the liquid's color and level.
4
5The new scene adopts a dark gray studio background, with the product placed on a low-reflection dark tabletop. There is a soft large-area light source on one side and a restrained warm rim light on the back.
6
7Focus on processing the optical relationship between the product and the new environment:
8The transparent area should see the new background, and cannot retain the background color or objects from the original photo;
9Present visually reasonable light transmission and refraction changes based on the product shape;
10Surface reflections should match the new light source;
11Retain natural contact shadows at the bottom.
12
13Allow reflections, lighting, and the background seen through the product to change, but do not change the product's own structure, color, labels, and contents.
14
15Do not turn transparent glass into opaque plastic, do not turn matte material into completely transparent, and do not add chrome effects to areas that didn't have them.
16
17Retain the original product's shooting angle, do not show the back without reference. Generate only one 3:4 vertical product photography image, no props or copy.

1Please generate a natural, believable real-life product display photo based on the two images I uploaded.
2
3Image 1 is the person reference, determining the person's identity, facial features, hairstyle, skin tone, and clothing.
4Image 2 is the product reference, determining the product's appearance, structure, color, material, Logo, and packaging information.
5
6Let the person in Image 1 naturally hold the product in Image 2, as if displaying it in front of the camera. The product is located in front of the chest or to the side of the body, with the main display side facing the camera, not blocking the person's face.
7
8Choose a reasonable holding method based on the product shape. Fingers should naturally wrap, hold, or pinch the product, with correct front-back occlusion, contact shadows, and force relationships. Hands cannot pass through the product, and the product cannot float above the palm.
9
10Retain the person's recognizable facial relationships, sense of age, hairstyle, and skin details; no face swapping, no changing body type. Allow arms and hands to adjust posture to pick up the product.
11
12The product cannot be redesigned. Retain the original outline, bottle cap or other structures, label positions, main color scheme, and recognizable text. Do not mirror the Logo, do not add accessories.
13
14Let the person and the product be in the same lighting set to avoid a stiff collage feel. The background follows the person's photo, appropriately simplifying distracting elements.
15
163:4 vertical, both person and product are clearly visible, no headlines, slogans, or purchase reviews added.

1Please synthesize the multiple product photos I uploaded into a unified shooting style combination main image.
2
3The main product in each reference image is treated as an independent and non-confusable object. Each product appears only once in the final image, no omissions, no repetitions, no merging.
4
5Accurately retain each product's own appearance, structure, color, material, Logo, labels, and recognizable text. Do not mix bottle caps, color schemes, packaging patterns, or labels of different products.
6
7Place all products on the same studio tabletop, forming a simple, layered combination display. There can be slight front-back relationships between products, but the main display side and brand information should be clearly visible.
8
9Try to follow the angles already shown in each product reference image; do not actively show backs lacking reference.
10
11Unify light direction, color temperature, clarity, and contact shadows to make them look like they were completed in the same shoot, avoiding a collage feel with different backgrounds and lighting.
12
13The background adopts a simple neutral color coordinated with the color scheme of the entire group of products, not overpowering the subject. Do not change product colors to match the background.
14
15Do not add unprovided products, gifts, gift boxes, packaging bags, prices, or promotional text.
16
17Generate a 1:1 square combination main image, with all products completely within the frame.

1This round only modifies: 【Target Change】.
2
3Must retain: 【Identity, composition, object position, text, color scheme, and other key items】.
4
5Allow linkages generated by modification: 【Necessary contact shadows, reflections, occlusions】.

1Image 1 is only used to determine the subject's identity and appearance.
2Image 2 is only used to reference color scheme, brushstrokes, and material texture.
3Image 3 is only used to reference composition and subject proportion.
4
5Do not bring the person from Image 2 or the objects from Image 3 into the final image.

1The image must feature the following text, retained verbatim, each item appearing only once:
2"【Title】"
3"【Subtitle】"
4
5The title is located at 【Position】, and the subtitle is located at 【Position】.
6Do not add other text.

1Adopt a restrained color scheme and clear visual hierarchy.
2Retain real material details and reduce decoration.
3Leave large areas of quiet space outside the subject.
4Present surface undulations through soft side lighting, avoiding over-sharpening and exaggerated highlights.

1First extract the clearest set of relationships in the theme, then translate it into a visible visual event.
2
3Use only one core metaphor per image.
4Let the subject, action, and spatial relationship bear the expression.
5Do not pile in multiple symbols at the same time, and do not rely on explanatory text to be understood.