Best AI Image? Midjourney V6 vs DALL E 3 vs Stable Diffusion
Vložit
- čas přidán 31. 12. 2023
- Are you looking for the best AI image generator? Midjourney V6 vs DALL E 3, vs Stable Diffusion.
In this video, we compare the three most popular AI image generators
Subscribe to stay up to date on future videos / @masteraifast .
For more Midjourney tips, tricks and comparisons • Midjourney
Check out the MasterAIFast Patreon!
/ masteraifast
#midjourney #dalle #stablediffusion - Věda a technologie
This Midjourney playlist offers guides to getting started, pro tips, creating consistent characters and more - czcams.com/play/PLQx9bk7yqXYgeK8ywdx0e40g4en-_-0lo.html
It seems that DALL-E 3 has quietly improved in terms of quality, without any notification or announcement to its users. This is different from Midjourney or Leonardo AI, which always announce their latest versions. I've just realized the recent quality improvement shown by DALL-E 3. And it seems like the intelligence of DALL-E 3 is connected with ChatGPT 4; as ChatGPT becomes smarter, so does DALL-E 3. Does that sound logical ?
The censorship and more importantly restrictions in dalle is really ruining for me. They are way too cautious in dalle
Dalle-3 is the better than all but hamstrings itself with censorship. That’s what I’m here looking for the next best AI image tool
I still like midjourney more just because the images are more realistic for my projects. But I also see potential on the other Ai apps.
Nicely done! Thank you! Perhaps in future comparisons like this, a short 2-second pause after asking your viewers to guess (without them having to physically pause the video) would be the only improvement I have to recommend.
All these comparison videos judge the different products in a way that is not fair or accurate. Just understanding the prompt is 1 critique, but each product requires different type of prompts to get the best results, you should also judge them at their best with the best prompts made for each.
So informative. I learned to look much closer at the images I see in front of me.
Wow these images are out of this world 🙌 Amazing video. So helpful! 🙏
This kind of "Prompt Off" is great. More of these please.
I love how this highlights that AI's are prone to generating stunning, pleasing images which often distracts us from the fact that they don't take art direction very well. If they were an apprentice or contract artist you'd say "Nice image, bro, but did you even read the spec?"
While a fun experiment, its a bit rough on stable diffusion. Dall-E and MJ are trained/designed to work on conversational prompts. SD does not perform well with that prompt style. It requires a programmatic style of prompt, and its quite tricky to tease out. But when you hit a good prompt its amazing.
Disagree. SDXL isn't as good at interpreting prompts. But it has many more fine tuning options which, when used correctly, produce an even higher quality image than the competition in my experience. It also does not account for image to image capabilities.
I like your way of talking but what value has just one image per prompt? For the next seed your ranking could be otherwise. I would recommend to do at least a test with some more images (10? 20?) and that you explain what model (e.g. in stable diffusion) and parameters you are using, as they differ a lot even on the same prompt.
Perfect score! That was fun and a valuable comparison. Thank you for putting that together. I’d love to see the same thing done with image to video animation. That is, Pika, Runway and one or two of the others.
I prefer realism and detail which makes midjourney my top choice. Dalle-E 3 seems too cartoonish for my taste.
Love your videos so much! Can you make a tutorial video on FlexClip’s AI tools? Really looking forward to that!🥰
Thank you!
I just find this video. I took the text and tried the text in Adobe Firefly, Facebook meta Ai and Microsoft copilot. Different results
I use all 3 , Dall E is great for drawings, but Mid is better at the rest, period!
Dalle is the best a drawings, especially sketches made to look like a talented human sketch. But you’re right Midjourney is better at everything else
Which model did you use with Stable Diffusion ? SD 1.5 ? SD 2.1 ? SDXL ? any custom ?
Hey! I used SDXL
you should redothe test with custom model or your test isn't fair at all@@MasterAIFast
@@thibaudherbert3144which custom model?
You could also judge witch AI bot in these comments is the best.
That first golden retriever photo killed me.
objctively midjourney images are the more realistic..however dal e better captures he prompt
What about meta imagine ai. That one is phenomenal. Its a 4 way competition now.
An incredible tutorial. Absolutely fantastic.
I have seen a couple of these comparisons and the winner varies. SD seems to be the worst of the 3 in this vedio. Its result depends on the used checkpoints, loras and extensions.
Fair point. This video is geared towards beginners who are looking for like for like comparisons. Agree adding loras, extensions, etc. could change the outcome.
Yes! Let us hail the biggest exploit of artists all over the world in the history of mankind.
Gross.
Skill issue
Stay mad and poor loser