Midjourney v5 - Style Prompt Tips and Reference Tricks
TLDRThe video discusses techniques for refining image outputs in Midjourney V5 using specific prompts and style references. It addresses common issues such as ignored instructions and aspect ratio challenges, offering solutions like prompt restructuring and the use of Leonardo's canvas feature for aspect ratio adjustments. The host, Tim, shares his experiments with various styles and artists, ultimately suggesting a combination of tools and methods for achieving desired results.
Takeaways
- 🎨 Midjourney V5 has improved its linguistic understanding, but still requires a mix of programming and literary language for effective prompting.
- 🖼️ The aspect ratio of the output can influence the composition, with 16:9 tending towards cinematic waist-up shots.
- 📸 The prompt 'In the style of' may not always yield the desired artistic style, as demonstrated with Frank Miller's style.
- 👗 The AI sometimes ignores specific details like the character being shoeless, possibly due to the training data.
- 🔄 Rearranging the prompt and using a formula (cinematic still, film by, scene, subject, action, set, shot) can help achieve closer results.
- 🌲 Using image prompts can lock you into the aspect ratio of the reference image, but there are ways to work around this.
- 🎭 Emoting or specific character actions can be challenging to capture; using photo bashing can help convey these emotions.
- 👥 The AI's understanding of styles is based on its training data; it may struggle with styles not commonly associated with the subject.
- 🔍 Experimenting with different artists and styles can lead to unexpected but interesting results.
- 🛠️ Combining various tools and techniques, like Leonardo's canvas feature, can help refine and expand the AI's output to desired formats.
Q & A
What is the main focus of the video?
-The main focus of the video is to provide tips on how to control the output of Midjourney V5, specifically when it comes to using posing and image references to get desired results.
What issue did Eric Schlitzbeyer raise about Midjourney?
-Eric Schlitzbeyer raised the issue that Midjourney often ignores specific instructions, such as wanting a full body picture and the addition of 'in the style of' which seems to be completely disregarded.
How does the video creator attempt to address Eric's concerns?
-The video creator attempts to address Eric's concerns by providing examples and demonstrating various prompt techniques to achieve closer results to the desired image, including adjusting the prompt structure and using different styles.
What is the significance of Frank Miller's style in the video?
-Frank Miller's style is significant because it is used as an example of a specific artistic style that the video creator tries to apply to the Midjourney V5 output, but finds it challenging to achieve.
What is the 'prompt formula' mentioned in the video?
-The 'prompt formula' mentioned in the video is a structured way of giving instructions to Midjourney V5, which includes elements like 'cinematic still', 'film by', 'scene', 'subject', 'action', 'set', and 'shot', to better control the output.
Why does the video creator use different artists and directors in the prompts?
-The video creator uses different artists and directors in the prompts to experiment with various styles and to see how Midjourney V5 interprets and applies these styles to the generated images.
What is the problem with using image prompts in Midjourney V5?
-The problem with using image prompts in Midjourney V5 is that the aspect ratio of the reference image locks the output aspect ratio, which may not always be desired, and Midjourney V5 sometimes insists on including certain elements like shoes, even if they are not specified in the prompt.
How does the video creator handle the aspect ratio issue in the generated images?
-The video creator handles the aspect ratio issue by using the canvas feature in Leonardo to adjust and expand the image according to the desired format, such as 16:9.
What is the result of combining Midjourney V5 and Leonardo?
-Combining Midjourney V5 and Leonardo allows the video creator to generate images, turn them into illustrations, and then use the canvas feature in Leonardo to adjust and expand them into the desired format, creating a unique blend of styles and compositions.
What is the video creator's conclusion about achieving specific styles in Midjourney V5?
-The video creator concludes that achieving specific styles in Midjourney V5 can be challenging, especially when the style is not directly associated with the subject matter, and that sometimes, the AI has difficulty applying certain styles to new subject matters.
Outlines
🎨 Exploring Prompting in Mid-Journey V5
The video begins with the creator addressing the audience about enhancing their understanding of prompting in Mid-Journey V5, particularly when it comes to obtaining specific images. The creator acknowledges the challenges in getting desired outputs and aims to provide tips for better control over the output. The video is inspired by a comment from Eric Schlitzbeyer, who points out the limitations of Mid-Journey in following detailed instructions and style preferences. The creator then shares an example prompt provided by Eric, which describes a scene involving a 10-year-old Viking girl in a threatening forest, and discusses the results generated by Mid-Journey based on this prompt. The creator also talks about the new prompting features in Mid-Journey V5 and shares their thoughts on its effectiveness. They then delve into a detailed explanation of how to improve the prompt to get closer to the desired image, using a specific formula and adjusting elements of the prompt.
🖌️ Refining Prompts and Style Experimentation
In this paragraph, the creator continues their exploration of prompting in Mid-Journey V5 by discussing the challenges of achieving the desired Frank Miller style and full-body shots. They experiment with different prompt modifications, including emphasizing Frank Miller's name and using a Viking warrior image as a reference. The creator also highlights the limitations of image prompts and aspect ratios, sharing a trick to overcome these restrictions. They further experiment with various artists' styles, such as Mike Mignola and Akira Kurosawa, and discuss the results obtained. The creator also shares their thoughts on Mid-Journey's training and its ability to stylize based on familiar themes versus more unique combinations. The paragraph concludes with the creator's attempts to capture the Viking girl's emotion through a photo manipulation technique.
🌟 Final Experiments and Conclusion
The final paragraph sees the creator conducting further experiments with the Viking girl prompt, exploring different styles and aspect ratios. They discuss the challenges of getting Mid-Journey to accurately portray a screaming war cry and share a technique for adding specific emotions to characters using photo manipulation. The creator then talks about their attempts to adjust the aspect ratio of the generated images from 2.33:1 to 16:9, highlighting the limitations of Mid-Journey in this area. They introduce Leonardo's canvas feature as a solution for expanding images to the desired format. The creator wraps up the video by encouraging viewers to share tips and questions in the comments and promotes their upcoming video on cinematic prompting in Mid-Journey. They conclude by thanking the audience for watching and expressing their hope for future growth of their channel.
Mindmap
Keywords
💡Midjourney V5
💡Prompting
💡Image References
💡Style
💡Aspect Ratio
💡Cinematic
💡Frank Miller
💡Viking Girl
💡Gloomy Mystical Threatening Forest
💡Image Prompts
💡Emoting
Highlights
The video aims to provide tips on controlling output in Midjourney V5 for specific image references.
Eric Schlitzbeyer's comment about Midjourney ignoring instructions inspired the video.
The challenge of getting a full body image and the style of Frank Miller was discussed.
The importance of linguistic prompting and natural language in Midjourney V5 is highlighted.
The video demonstrates how to adjust prompts to achieve closer results to desired images.
The prompt formula used in cinematic Midjourney is introduced.
The issue of Midjourney's tendency towards cinematic compositions and waist-up shots is discussed.
An example of how to modify the prompt to get full body shots and the desired Frank Miller style is given.
The impact of adding 'Sin City' to the prompt and its results on the generated images is shown.
The strategy of emphasizing certain words in the prompt by using 'colon colon' is explained.
The use of image prompting with Midjourney and its limitations is explored.
A trick to change aspect ratios using Leonardo's canvas feature is demonstrated.
Experiments with different artists' styles, such as Mike Mignola and Akira Kurosawa, are discussed.
The problem of Midjourney's difficulty in stylizing unfamiliar subjects is highlighted.
The creative process of combining various prompts and tools for better results is encouraged.
A technique for adding specific emotions to characters using photo bashing is introduced.
The video ends with a call to action for viewers to like, subscribe, and support the channel.