Stable Diffusion is one of the popular AI-based image generators out there and serves its users as a free alternative to Midjourney.
Stable Diffusion is not as advanced as Midjourey, but if you are using the prompt with appropriate details and information, you will get results that look as good as Midjourney’s. It won’t be wrong to say that a prompt is a significant ingredient in creating a good image using AI tools.
Looking at the images generated by the community, you must be wondering about the prompts they used to create an image like that. Well, Reverse Image Prompt Generator is all you need to determine the input prompt. The suggested prompt can be tweaked a bit to get the best results.
Here, in this post, you will learn about the best reverse image prompt generator when we pit two popular tools against each other.
Reverse Image Prompt Generator for Stable Diffusion
The text description can be turned into an image using tools like Stable Diffusion. But, the reverse of the same is possible as well.
A small change in the text description or the prompt submitted to an AI tool can make a lot of difference. If you are using the right prompt, it is possible to get the best results from tools like Stable Diffusion.
There are tools called Reverse Image Prompt Generator that can suggest the prompt based on the image you provided. And there are many options available in the market. But, here, we will review the two popular Reverse Image Prompt Generators – Img2prompt and Clip Interrogator.
Procedure for Review
In this post, we will compare the prompts generated by these tools and find out which works best.
For the testing, we generated 7 images using ready-made prompts; we called them Original Prompts. And the images generated by the Original Prompts are called Original Images.
Next, we uploaded these Original Images on Img2prompt and Clip Interrogator as input and recorded the output – Suggested Prompt.
Finally, the suggested prompts by both tools were used to generate new images to see if they had the visual qualities as the original one.
The Suggested Prompts could have been used to create better images, but we fed the output of both Reverse Image Prompt Generators in Stable Diffusion without any changes and only once.
Test 1 – Model and Painting style identification
Original prompt –
a painting of a thinker no facial hair, thoughtful, focused, visionary, calm, jovial, loving, fatherly, generous, elegant well fed elder with few eyebrows and his on from Kenya by Henry Ossawa Tanner . dramatic angle, ethereal lights, details, smooth, sharp focus, illustration, realistic, cinematic, art station, award-winning, rgb , unreal engine, octane render, cinematic light, macro, depth of field, blur, red light and clouds from the back, highly detailed epic cinematic concept art CG render made in Maya, Blender and Photoshop, octane render, excellent composition, dynamic dramatic cinematic lighting, aesthetic, very inspirational, arthouse
Original image –
Prompt Suggested by Clip interrogator –
a painting of an old man with a mustache, saadane afif, robert crumb photorealism, michael shapcott, expressive impressionist style, jemal shabazz, gandhi, 16:9, published art, terracotta, watercolor technique, scott adams, mid – 3 0 s aged, tim booth, color study, mid – 30s, \’the soul creates
Result using the prompt –
Prompt Suggested by Img2prompt –
a painting of an old man with a mustache, an ultrafine detailed painting by Philip Evergood, behance contest winner, figurative art, detailed painting, hyperrealism, oil on canvas
Result using the prompt –
Review – (Img2prompt – 1 vs Clip Interrogator – 0)
For this particular image, Clip Interrogator failed to identify the painting style, that is, ‘Oil Painting,’ and identify it as ‘Watercolor’ style. Moreover, the image produced are not very impressive; most of them are distorted. Users will have to provide some negative prompts as extra input.
Img2prompt produces acceptable images. The color technique used is appropriate.
Test 2 – Minimalism and details identification
Original prompt –
interior design, open plan, kitchen, and living room, modular furniture with cotton textiles, wooden floor, high ceiling, large steel windows viewing a city, 35mm camera, realistic, 8k
Original image –
Prompt Suggested by Clip interrogator –
a living room filled with furniture and a wooden table, splendid haussmann architecture, brooklyn, pexels contest winner, inspired by Toros Roslin, beams, kitchen counter, beautiful terrace, rundown new york apartment, trending on interfacelift, german renaissance architecture
Result using the prompt –
Prompt Suggested by Img2prompt –
a living room filled with furniture and a wooden table, a digital rendering by Paul Georges, trending on pinterest, maximalism, maximalist, minimalist, made of wrought iron
Result using the prompt –
Review – (Img2prompt – 1 vs Clip Interrogator – 1)
In this case, both Img2prompt and Clip Interrogator produces images of the same quality, but Img2prompt failed to understand the ‘realistic’ element in the original image.
Clip Interrogator is the clear winner here.
Test 3 – Scene depth identification
Original prompt –
dark and terrifying horror house living room interior overview design,, Greg Rutkowski, Zabrocki, Karlkka, Jayison Devadas, Phuoc Quan, trending on Artstation, 8K, ultra wide angle, pincushion lens effec
Original image –
Prompt Suggested by Clip interrogator –
a living room filled with furniture and a chandelier, unreal engine. film still, still from horror movie, vray beautiful, hyper real acrylic painting, anamorphic flares, buildings photorealism, atmospheric red effects, very realistic. low dark light
Result using the prompt –
Prompt Suggested by Img2prompt –
a living room filled with furniture and windows, a raytraced image by Christian W. Staudinger, cg society contest winner, photorealism, vray tracing, volumetric lighting, vray
Result using the prompt –
Review – (Img2prompt – 1 vs Clip Interrogator – 2)
Here’s another image for a living room. But, as you can see in the original image, a red-shade element gives the image a scary look. Img2prompt, again, failed to understand the required depth. Clip Interrogator, on the other hand, identifies and describes the original image as ‘still from horror movie’ having shades of red.
Yet again, Clip Interrogator is the clear winner here!
Test 4 – Human model, Camera type, and background identification
Original prompt –
professional portrait photograph of a gorgeous Norwegian girl in winter clothing with long wavy blonde hair, ((sultry flirty look)), freckles, beautiful symmetrical face, cute natural makeup, ((standing outside in snowy city street)), stunning modern urban upscale environment, ultra realistic, concept art, elegant, highly detailed, intricate, sharp focus, depth of field, f/1. 8, 85mm, medium shot, mid shot, (centered image composition), (professionally color graded), ((bright soft diffused light)), volumetric fog, trending on instagram, trending on tumblr, hdr 4k, 8k
Negative: (bonnet), (hat), (beanie), cap, (((wide shot))), (cropped head), bad framing, out of frame, deformed, cripple, old, fat, ugly, poor, missing arm, additional arms, additional legs, additional head, additional face, multiple people, group of people, dyed hair, black and white, grayscale
Original image –
Prompt Suggested by Clip interrogator –
a woman posing for a picture in the snow, glorious long blong hair, realistic, shaded perfect face, beautiful illumination, evening makeup, 2019 trending photo, by John Clayton, stylized portrait h 640, streaming on twitch, elsa from frozen, nordic, cute beautiful, winter night, contourless
Result using the prompt –
Prompt Suggested by Img2prompt –
a woman with long blonde hair wearing a blue jacket, a photorealistic painting by Alexander Kucharsky, featured on cg society, art photography, white background, pretty, uhd image
Result using the prompt –
Review – (Img2prompt – 1 vs Clip Interrogator – 3)
Competition gets tricky in this case. Clip Interrogator carefully understood the elements in the original image, like the bokeh mode produced by the camera lens. The background in the original image is identified correctly as ‘snowy.’
Img2prompt produced good images as well; the character identification is correct. However, the overall prompt is not detailed as Clip Interrogator.
Clip Interrogator gets another point.
Test 5 – Image art and details identification
Original prompt –
a highly detailed matte painting of a man on a hill watching an alien spaceship launch in the distance by studio ghibli, makoto shinkai, by artgerm, by wlop, by greg rutkowski, volumetric lighting, octane render, 4 k resolution, trending on artstation, masterpiece
Original image –
Prompt Suggested by Clip interrogator –
a man sitting on top of a lush green hillside, ralph mcquarrie. centered image, anamorphic lens flare, inspired by Michał Karcz, space graphics art in background, cave background, psychedelic artwork, outside in space, attack on titan scenery, space in background
Result using the prompt –
Prompt Suggested by Img2prompt –
a painting of a man sitting on a hill looking at a distant object, a matte painting by David A. Hardy, cgsociety, fantasy art, matte painting, sense of awe, sci-fi
Result using the prompt –
Review – (Img2prompt – 1.5 vs Clip Interrogator – 3.5)
Clip Interrogator and Img2prompt did a satisfactory job of suggesting prompts that generate similar images. However, none of them identifies the sci-fi background element like a spaceship. But, we believe this can be added manually when generating images.
Well, it all boils down to preference, half points to both.
Test 6 – Funko model identification
Original prompt –
full body 3d render of a girl blonde, Argentina national football team member, as a funko pop! , studio lighting, white background, single body, no shadow, blender, trending on artstation, 8k, highly detailed
Original image –
Prompt Suggested by Clip interrogator –
a close up of a person holding a soccer ball, as a full body funko pop!, lady kima, motion design, platinum blonde long hair, inspired by Kaff Gerrard, gif, merry england, gameplay footage, early screen test, full height view, white uniform, photoscan, funko
Result using the prompt –
Prompt Generated by Img2prompt –
a girl in a soccer uniform holding a soccer ball, a character portrait by Louisa Puller, reddit contest winner, figurativism, made of plastic, hyper-realistic, ultra realistic
Result using the prompt –
Review – (Img2prompt – 1.5 vs Clip Interrogator – 4.5)
Clip Interrogator is a clear winner here. As you can see Img2prompt couldn’t identify that the figure is a Funko Pop.
Test 7 – Painting type and artist identification
Original prompt –
A beautiful portrait of a girl with blonde hair by Rembrandt
Original image –
Prompt Suggested by Clip interrogator –
a painting of a young girl wearing a white hat, rembrandt style painting, radiohead singer thom yorke, frill, rubens, artforum, google arts and cultures, punk little girl, trending on artforum, windblown, cascade helmet, short curly blonde haired girl, cone, young boy, greta thunberg, gilt
Result using the prompt –
Prompt Suggested by Img2prompt –
a painting of a young girl wearing a white hat, a flemish Baroque by Cornelisz Hendriksz Vroom, the Younger, Artstation, art informel, flemish baroque, dutch golden age, rococo
Result using the prompt –
Review – (Img2prompt – 2 vs Clip Interrogator – 5)
The quality of the images produced by Stable Diffusion doesn’t always depend on the prompt input. You might get better images using the same prompt at different sessions. Images generated by Img2prompt look good here but need more primary elements: the artist’s painting style.
Clip Interrogator identified ‘Rembrandt’ painting style, but the prompt returned an error code because of using Swedish Activist Greta Thunberg’s name. As you may already know, Stable Diffusion has banned using a set of words, and using names of famous personalities is also banned.
The image generated using Clip Interrogator’s suggested prompt neatly depicts Rembrandt’s painting style.
Conclusion
Clip Interrogator is the best Reverse Image Prompt Generator you can use. The tool identifies all the essential elements of the input image and mentions the same in the suggested prompt.
Remember to upload a clear image first to get the best out of Clip Interrogator. The suggested prompt won’t be the perfect outcome, and you will need to tweak it a bit. Start by deleting the duplicate instructions and details in the prompt, then add extra required details.
And that’s pretty much it!
