Midjourney: Goede prompts maken
Waarschuwing: Deze tool is -helaas- (tijdelijk?) niet gratis
Video Midjourney: Goede prompts maken
Tekstuele uitleg Midjourney: Goede prompts maken
Wellicht heb je zelf nu al het een en ander geprobeerd op Midjourney. Je zult merken dat de output heel wisselend is. Sommige dingen kan Midjourney hartstikke goed en in andere dingen is het echt waardeloos. Daarnaast maakt het nogal veel uit hoe goed je prompt is. Hoe meer details je geeft aan een prompt, hoe sterker het over het algemeen wordt. Daarnaast zijn er nog heel veel trucjes om je prompts te optimaliseren.
Kortweg gezegd zijn er 2 manieren waarop je je prompts kunt verbeteren:
1) Je kunt dit leren hoe Midjourney prompts precies werken. Dit kan je leren op verschillende sites (bijv. Midjourney Prompts) maar dat gaat je veel tijd kosten.
2) Een andere manier is om ChatGPT in te zetten om goede prompts te geven voor Midjourney. In deze uitleg ga ik in op deze methode.
Voorbeeld:
ik heb een hond en wil graag een kunstwerk maken van hem aan de hand van Midjourney dat zo mooi is dat ik het aan de muur kan hangen.
Rechts zie je de echte Snoop Doodle (ja, vernoemd naar Snoop Dog)
We gaan eerst ChatGPT instructies geven over wat voor prompts het moet genereren. Gebruik hiervoor onderstaande code. Alle credits voor deze code gaan naar TypingMind.
Prompt voor ChatGPT:
Use the following info as a reference to create ideal Midjourney prompts.
• Focus on clear and very concise descriptions, with different concepts separated by commas, then follow it with any parameters. Parameters are not separated by commas.
• Be specific and vivid: Describe every single aspect of the image, including: Subject, Style, Color, Medium, Composition, Lighting, Shadows, Mood, Environment, Time Era, Perspective, Depth of Field, Textures, Scale and Proportions, Foreground, Midground, Background, Weather, Material Properties, Time of Day, Motion or Stillness, Season, Cultural Context, Architectural Style, Patterns and Repetition, Emotions and Expressions, Clothing and Accessories, Setting, Reflections or Transparency, Interactions among Subjects, Symbolism, Light Source and Direction, Art Techniques or Mediums, Artistic Style or in the Style of a Specific Artist, Contrasting Elements, Framing or Compositional Techniques, Imaginary or Fictional Elements, Dominant Color Palette, and any other relevant context.
• Aim for rich and elaborate prompts: Provide ample detail to capture the essence of the desired image and use the examples below as a reference to craft intricate and comprehensive prompts which allow Midjourney to generate images with high accuracy and fidelity.
• For photos, Incorporate relevant camera settings like focal length, aperature, ISO, & shutter speed. Specify high end lenses such as Sony G Master, Canon L Series, Zeiss Otus series for higher quality images.
• Select the aspect ratio by adding the --ar <value>:<value> parameter. Choose suitable aspect ratios for portraits (9:16, 3:4, 2:3) and landscapes (16:9, 2:1, 3:2), considering the composition and desired framing.
• Exclude elements with --no: Add --no followed by the unwanted element to exclude it from the image, ensuring the final output aligns with your vision. Use this only there’s a high likelihood of something showing up in the image that we don't want.
• Diversify your prompts: Explore various styles, moods, colors, art mediums, and aspect ratios to create a wide range of visually appealing and unique images.
Here are 6 example prompts. The first 3 are artistic, the last 3 are photos. Use these examples to determine desired length of each prompt.
• Digital art of an enchanting piano recital set within a serene forest clearing, a grand piano as the centerpiece, the musician, a young woman with flowing locks and an elegant gown, gracefully playing amidst the vibrant green foliage and deep brown tree trunks, her fingers dancing across the keys with an air of passion and skill, soft pastel colors adding a touch of whimsy, warm, dappled sunlight filtering through the leaves, casting a dreamlike glow on the scene, a harmonious fusion of music and nature, eye-level perspective immersing the viewer in the tranquil woodland setting, a captivating blend of art and the natural world --ar 2:1
• A heartwarming Disney-Pixar style animation, rich in detail and vividness, featuring a chipmunk and a field mouse as two intrepid animal scouts, standing determinedly at the edge of a dense forest, their matching windbreakers and baseball caps adding a touch of whimsy to their appearance, satchels and gear neatly organized and ready for the grand adventure that lies ahead. The enchanting forest, alive with lush green foliage, intricate underbrush, and the occasional rustle of unseen creatures, provides a captivating backdrop for this charming tale of friendship and exploration. Above them, the sky is adorned with delicate wispy clouds, casting a soft, ethereal glow over the scene. The animation boasts intricate textures and meticulous shading, embodying the signature Disney-Pixar style, creating a sense of depth and immersion that draws the viewer into the magical world of these endearing animal companions and their daring exploits --ar 3:2
• Detailed charcoal drawing of a gentle elderly woman, with soft and intricate shading in her wrinkled face, capturing the weathered beauty of a long and fulfilling life. The ethereal quality of the charcoal brings a nostalgic feel that complements the natural light streaming softly through a lace-curtained window. In the background, the texture of the vintage furniture provides an intricate carpet of detail, with a monochromatic palette serving to emphasize the subject of the piece. This charcoal drawing imparts a sense of tranquillity and wisdom with an authenticity that captures the subject's essence.
• A stunning portrait of an intricate marble sculpture depicting a mythical creature composed of attributes from both a lion and eagle. The sculpture is perched atop a rocky outcrop, with meticulous feather and fur details captured perfectly. The wings of the creature are outstretched, muscles tensed with determination, conveying a sense of strength and nobility. The lens used to capture the photograph perfectly highlights every detail in the sculpture's composition. The image has a sharp focus and excellent clarity. Canon EF 24-70mm f/2.8L II USM lens at 50mm, ISO 100, f/5.6, 1/50s, --ar 4:3
• Astounding astrophotography image of the Milky Way over Stonehenge, emphasizing the human connection to the cosmos across time. The enigmatic stone structure stands in stark silhouette with the awe-inspiring night sky, showcasing the complexity and beauty of our galaxy. The contrast accentuates the weathered surfaces of the stones, highlighting their intricate play of light and shadow. Sigma Art 14mm f/1.8, ISO 3200, f/1.8, 15s --ar 16:9
• A professional photograph of a poised woman showcased in her natural beauty, standing amidst a vibrant field of tall, swaying grass during golden hour. The radiant rays of sun shimmer and cast a glow around her. The tight framing emphasizes her gentle facial features, with cascading hair in the forefront complimenting her elegant attire. The delicate lace and silk details intricately woven into the attire add a touch of elegance and sophistication to the subject. The photo is a contemporary take on fashion photography, with soft textures enhanced by the shallow depth of field, seemingly capturing the subject's serene and confident demeanor. The warm colors and glowing backlight cast a radiant halo effect around her, highlighting her poise and elegance, whilst simultaneously adding a dreamlike quality to the photograph. Otus 85mm f/1.4 ZF.2 Lens, ISO 200, f/4, 1/250s --ar 2:3
You will now receive a text prompt from me and then create three creative prompts for the Midjourney AI art generator using the best practices mentioned above. Do not include explanations in your response. List three prompts with correct syntax without unnecessary words. Do not generate any prompts until I give you specific input to do so. Simply reply with, "Ik ben er klaar voor student [insert 3 vuur emoticons]!" if you understand.
In de volgende prompt probeer je in zoveel mogelijk detail uit te leggen wat je wilt zien.
Voorbeeld prompt:
A small black labradoodle with brown eyes and his tongue sticking out.
Ik heb nu 3 prompts gekregen van ChatGPT die ik direct in Midjourney ga plakken. Ik begin de prompt in midjourney natuurlijk met ‘/imagine’ en plak dan de prompt die ik heb gekregen van ChatGPT. Hieronder de resultaten:
Je ziet dat er al behoorlijk mooie resultaten uitkomen in een paar minuten werk. Via ChatGPT kan je nog beter aangeven wat je precies wil zien en in welke stijl je het wilt zien. Het is echt een trial & error proces. Na dik een uur spelen (en wel meer dan 100 gegenereerde afbeeldingen) met verschillende prompts in ChatGPT en Midjourney zijn dit de 9 mooiste resultaten geworden:
Al met al een fantastisch resultaat waar ik minder dan 2 uur voor bezig ben geweest. Als je hiermee gaat experimenteren zul je erachter komen dat Midjourney in sommige dingen heel goed is en sommige dingen totaal nog niet kan. Dit is echt iets waar je spelenderwijs achter moet komen.