The creation of sword-wielding characters in Midjourney has proven to be difficult. You might know that depicting hands and swords is the AI's weak point, which makes this task all the more difficult. It can be difficult to understand the nuances of having a character hold a sword. This problem is significantly reduced with the Niji5 version, which is why I prefer it. However, dynamics may have changed with the release of version 5.2. Today I'll put it to the test.
We will start by exploring a simple prompt to build a solid understanding between us and the AI. We want to make it easier to understand the requests asked to the AI by engaging with these basic ideas, resulting in interactions that are more efficient and precise.
A ninja holding a sword:
V5.2
prompt: a ninja holding a sword --ar 16:9
This fascinating image is an excellent surprise and emanates an exceptionally exceptional quality that is genuinely admirable. The character who expertly wields the sword is masterfully brought to life by the AI, demonstrating how well it understood our directions. The sword is perfectly placed in the character's hand, displaying wonderful attention to detail. A sense of authenticity and expertise is conveyed by the character's and their weapon's flawless integration. This outstanding result emphasizes the AI's capacity to interpret and adapt our requirements into a visually engaging picture, displaying its skill and artistic prowess.
Niji 5
prompt: a ninja holding a sword --ar 16:9 --niji 5
Although this image has some appealing characteristics, it is not without flaws. Its overall impact is diminished by a number of obvious failings. The sword's extended pommel throws off the aesthetic balance and lessens the intended visual charm. Additionally, the dragon's oddly formed nose seems out of place and creates a sense of discord in the composition. These flaws limit the image's potential, highlighting the significance of perfecting particulars to provide a more unified and alluring visual depiction. The image's quality would surely increase if these problems were resolved, allowing it to more effectively express the intended story and visual attraction.
Version 5.2 clearly wins the first round with an unquestionable score of 1-0. It's important to mention that the created image is completely random. If the goal is to create a specific image, a more precise prompt might be required. The current prompt emphasizes the unpredictable nature of its creative process.
A ninja holding a sword walking on a street:
V5.2
prompt: a ninja holding a sword walking on a traditional japanese village street, epic portraiture --ar 16:9
Although the settlement and character remain to be of a modest caliber and lack any truly extraordinary aspects, the image has taken on a more lifelike quality. But the most important thing is that he isn't holding a sword. Overall, the created image is really disappointing.
Niji 5
prompt: a ninja holding a katana walking on a traditional japanese village street, epic portraiture --ar 16:9 --niji 5
The character in this picture is carrying four swords, and holding two instead of just one. But each of them is flawed and unacceptable. Without those swords spoiling everything, this image would have been good and full of details.
In the end, neither of the two was able to generate the desired image. It could only demonstrate what it does best, which is producing entirely random illustrations. The test highlights AI's existing limitations by reiterating how difficult it is for it to create hands and swords. Version 6 is expected to be a big improvement, promising more freedom and improved capabilities. I excitedly await the chance to put it to the test on your behalf and report my findings to you.
Wallpapers
Social Plugin