Hyper-Real Prototyping with Speed and Control - AI for Industrial Design
ฝัง
- เผยแพร่เมื่อ 8 เม.ย. 2023
- How did Bryce and Bing Jun get Midjourney to create restrained and beautiful aesthetics in detailed watches and expressive furniture forms? The secret lies in a clever way to choose and combine input images of certain characteristics. Find out exactly what these parameters are in this video.
The AI for Industrial Design series is created by students exploring AI's capability in a semester course - Everything with AI - directed by Donn Koh at the National University of Singapore, Division of Industrial Design.
See more at industrialdesign.ai/
#industrialdesign #industrialdesigner #midjourney #midjourneyai #midjourneytips #chatgpt #artificialintelligence #stablediffusion #designeducation #idsa #nationaluniversityofsingapore #uxdesign #uidesign
I've been looking for a workflow like this and your video really helped. I work in furniture design and find that 99% of what midjourney makes is nearly impossible to produce and your workflow addresses this perfect. Thank you!
this is exactly the type of discussion and exploration on AI as a prototyping tool I've been having a really hard time finding online. Thanks for the video and keep it up!!
Excellent, thank you for sharing!
God bless you guys, this is insanely helpful, and you share it forward just like that. Love it!
Hi Bing Jun, nice video. Gerald here.
Damn, you guys are awesome. Thanks for doing this research.
Wow, thank you both for these amazing and helpful insights. I do have a queston regarding using sketches as a reference image. Did you try uploading an original concept sketch or would this be a completely different use case? Thank you again.
안녕하세요. 흥미롭게 잘 봤습니다. 어떤 Ai를 사용했나요? 저도 레퍼런스디자인과 DNA디자인 2개를 섞어 보고 싶어요.
Which website you are using for creating AI images
On what software do you exactly design all these?
Thank you.
Awesome episode guys. I’ve been playing around a fair bit with blend prompts and find the results much better than text only prompts. Question: in the watch example were you combining two image prompts (1 the watch & 2 the dog graffiti) in combination with textual prompt? Or were you using just one image + textual prompt?
Hi Ben
Great question! I was using both the watch and illustration image as image references. They form a good foundation for AI to create a clean image. With v5 on Midjourney, the image weight that I used was the minimum (--iw 0.5), and let prompts control the CMF and type of watch. So, if I wanted different watches, such as an elegant dress watch or a rugged field watch, the output would be adjusted as such. Finer details would be up to the designer to remix and push.
Hope this clarifies! -Bing Jun
Could you run through a couple examples of actually operationally going through the process? aka a tutorial?
are you using the blending prompt to create these DNA iterations in MidJourney?
You should rename your video to "Hyper-Real Concept Images with Speed and Control" - Because nothing shown here is a prototype which is a physical version. Also, As a professional IDer, I would also be careful as much of the concept furniture shown at beginning are not manufacturable at all. Ai can be a helpful tool to come up with inspiration, but its probably 5% of a long design process that still needs to happen. If you showed up for an interview with those as ideas, I would question you on manufacturability and why you went down those paths. Good luck guys-