Great comparison. I've been using MidJourney for about a year now. I've generated close to 40,000 images. As someone who is a perfectionist and expects depth and granularity from an AI art generator, I have to say that I have a strong preference for MJ. DALL-E is like Corel Draw where as MJ is like Photoshop. These image generators are both mind blowing and deserve due credit for the monumental accomplishments they're making.
I started doing this last April using Bing at first, then Night Cafe. I avoided Midjourney for several months just because of discord. I like doing more artsy stuff but loved Midjourney 5.2 for everything else. Dalle-3 refuses to do anything that could be called artsy, and Midjourney 5.2 had some limitations in that area. But Midjourney 6 is amazing. The only area where Dalle-3 beats Midjourney 6 is with complex subject orientation, but you have to work with ChatGPT to really get the benefit from that. I do love using ChatGPT for writing prompts, and I have built some custom GPTs, but Dalle-3 is not reason enough to justify having a plus account. I am using Midjourney 6 for over 95% of my artwork now.
I would say that MJ is much better when it comes to photorealism, but I would give Dall-e the edge in illustration. One big difference between the two is that Dall-e takes the user's prompts very literally, and MJ tries to interpret the user's intent. A good example was the Pixar style robot image. While he preferred the MJ image, it really wasn't what he asked for in the prompt. He asked for the image to be reminiscent of Pixar animation style, and the Dall-e image was far more Pixar-like. It looks as though it could be a still from a pixar movie. The character is cute and marketable. The background, in particular, is much more like what you'd see as a background scene in one of their movies. He also asked for the robot to be shown in a playful manner, and the setting to be bright and cheerful. Dall-e followed those prompts, whereas the MJ image had a more serious tone with more muted colors. Merely comparing image to image, I like the style of the MJ creation over the Dall-e alternative, but when comparing the images to what the prompts asked for, MJ strayed while Dall-e stuck to them. As I said, when it comes to photorealism, MJ is the hands down winner, but when it comes to illustrations, Dall-e gives you what you ask for. He seemed impressed by the rust and weathering of the MJ robot, but that was an MJ decision, it wasn't in the prompts.
Hello, there are a good number of new videos on these best-known AI art generators almost on a weekly basis, which is occasionally overwhelming. While most of the video content of the other guys are filled with some humorous elements in an attempt to be engaging, there is always this element of depth and richness in your videos which makes the videos more informative and enriching. I also appreciate your quite fine sense of art when you comment on the images generated by the competitive AI generators -and I can't help but smile sympathetically when you incline in favor of Midjourney in regards to its peculiar approach (or philosophy as you define) to the image generation. If I'm not mistaken, you aim to reach out a certain group of people who do take the desire for professionalism seriously and strive for enriching themselves new and genuine information in the field. If this is true, I could tell you that you're exceptionally informative and enriching for them. I personally thank you for the good quality of content in your videos. Greetings (and sincere sense of thankfulness) from Turkey.
🎯 Key Takeaways for quick navigation: 00:56 🔄 *Mid Journey excels in changing parameters, fine-tuning algorithms, and managing works in its gallery. DALL•E operates mainly in chat GPT form with a standalone website version.* 02:57 🤖 *Mid Journey lacks an API, while DALL•E has an available API for users.* 03:11 🎨 *In a prompt battle, Mid Journey tends to offer more refined details and nuanced outputs, while DALL•E produces more saturated and broader-colored images.* 05:58 🧩 *Mid Journey outperforms DALL•E in creating seamless geometric patterns for elegant wallpapers.* 06:11 🖌️ *When designing logos, Mid Journey wins for better incorporation of diverse visual elements, although improvements are suggested for both platforms.* 07:20 🤖 *In the Pixar animation style, Mid Journey provides a more characterful and emotive robot, while DALL•E's output is more saturated and feels slightly childish.* 08:48 🎨 *Mid Journey's AI influencers appear more realistic and natural compared to DALL•E, which tends to produce oversaturated and less realistic images.* 13:12 🐉 *In an illustrated fantasy scene with a dragon, DALL•E adheres more closely to prompt specifics, but Mid Journey's composition is considered more interesting.* 14:45 🎌 *Mid Journey's tattoo design of a Phoenix in Japanese style feels more authentic, while DALL•E's version has a more graphic and symmetrical style.* 16:49 🦊 *In anthropomorphic animal illustrations, DALL•E tends to produce whimsical and childish images, while Mid Journey creates more dramatic and emotive scenes.* 18:25 🌌 *Mid Journey's surreal landscape with floating islands is preferred for its balanced composition and visual integrity over DALL•E's more scattered approach.* 20:58 🕍 *In detailed Gothic Cathedral illustrations, both Mid Journey and DALL•E showcase strengths, with Mid Journey offering a more atmospheric feel.* 23:36 🏠 *Mid Journey's approach to futuristic interior design prompts is refined and demonstrates thoughtful composition, while DALL•E tends to be more saturated and less considerate of overall composition.* 24:16 🤖 *Darly's unique strength lies in collaboration with ChatGPT, allowing interactive design discussions and iterative improvements.* 26:23 🕒 *Mid Journey's processing speed is comparable to Darly, and recent updates make it 2.7 times faster than the original V6 version.* 27:17 🚫 *Both platforms have content limitations, with Darly being more censored, restricting images of famous people or using artists as inspiration.* 28:38 💬 *Mid Journey boasts a more active and supportive community on Discord compared to Darly's evolving community.* 29:32 💲 *Pricing varies; Darly comes with ChatGPT Plus for $20/month with unlimited images, while Mid Journey starts at $10/month with GPU time restrictions and offers higher tiers for more features and GPU time.* 30:52 🎨 *Mid Journey leans towards Ultra realism and artist control, while Darly excels in specific object relationships, Illustrated, or painted styles and benefits from collaboration with ChatGPT inside ChatGPT Plus.* Made with HARPA AI
The whole Midjourney community and the company itself, are easily the best. They have excellent support, seem to be very transparent, spend lots of time on what matters, the users input etc....
Both Mj and DALL-E have distinct 'styles' IMO, a lot of the time you can tell by looking at the images which tool they were generated in (though not always). I suppose that is just the way with AI tools at the minute.
Was going to run these through Firefly, but didn't see the link for the prompts. Would love to see how Firefly stacks up against MJ and Dall-E. Of the two, definitely prefer MJ...by far! Edit: typed first prompt into Firefly and it more closely resembles MJ; it's actually quite good and stacks right up with MJ.
Thanks for creating this great video! I’ve been using MJ5, just started using 6, the coherence of MJ is frustrating but I’m hopeful it’ll improve, the quality of the imagery is just incredible! And it’s improved with v6! I like your prompt style, very descriptive and detailed, makes me imagine it in my mind’s eye! Thanks
Thanks for your video comparison. However, at this stage it's a little unfair to even put them in the same building. MJ is The Louvre. Dall-E is a porta-loo in The Louvre car park. Unfortunately someone has pushed it over and the contents are spilling out onto the street.
get both. dalle is amazing for illustrations and for refiining what you want over many generations with the help of gpt4, but it cant hold a candle to mdjourneys photorealism and pure splendor. its worth the combined 50 or 30 a month!
Videos like this are very important going forward.. it proves what human is.. how humans analyze, thank you for stepping up and taking on the challenge, you really did a good job, so midjourney wins this but dall e 3 is somewhat free and that means latent spaces take more time to interpolate In dall e 3
Blending Midjourney's artistic flair with DALL-E 3's realism enhances content creation. Midjourney excels in abstract, creative imagery, adding a unique creative touch, while DALL-E 3 specialises in lifelike detail, enriching the visual narrative. This combination offers diverse visual appeal, appealing to enthusiasts of AI, digital art, and practical content creation. It's a valuable technique for creators aiming to elevate their digital artwork and storytelling, merging the strengths of both AI tools for more compelling and varied visual content in RU-vid videos and blogs.
@@Kombo-Chapfika While Midjourney might not scream "unique" to everyone, it's got its own flair. It's like comparing apples and oranges with Darly - both have their pros and cons. Midjourney's knack for detail and mood sets it apart, but hey, Darly's not far behind with its vivid colors and patterns. It's all about what catches your eye!
Great! I like both, and I often ask GPT for ideas and I compare prompts from GPT in both DALL-E and Midjourney. Furthermore I'm starting to use Stable Diffusion through Pinokio
Did seemed to me that Midjourney was going to get the palm leaf whatever it did, I am left questioning how neutral this appraisal was. In the robot example, for example, Midjourney completely ignored most of the brief - something it is very prone to doing. Just like a real graphic artist then ...
Hey Samson! would love if you came on the MJ server and played an actual game of Prompt battles with the gang, We've been doing it since the start of Midjourney! My name on there is KinkyKong, and i'm one of the Green names in the guide section. I host a lot of the events on Midjourney. would love to colab on something!
However you can get either image - and both types of image with the proper prompt - you would simply edit one prompt to make it more comparable to the other - we are just looking at an interpretation, I would follow up with , worn and rust but full of heart, highly textured surroundings and so on..
We would like to get more functional with AI Image generators, but we are struggling to create a consistent character within. How to have the same character for different scenes? which platform can do this job the best and with the least complications?
So in Dall e and mj you can get consistent characters in mj you can use the previous images and use vary region. In dall e if your prompt is refined well it will always create the same image and you can use the gen-id to create that image in multiple scenes ect…
That's a superb review of the apps and a thorough comparison. I share your thoughts, on both these apps. I use them at work daily. I use SD to upscale and refine the images from MJ for use in my advertising projects. A colleague is urging me to give a go at Comfy UI, which looks smashing for some examples he produced. I find the process laboured, and the output misses what I get in MJ, everytime I try to use it. What are your thoughts on that?
I would say that MJ did a lot better copying the best artworks of real humans, whereas dall-e tries harder to be more original, but thus creating less beautiful / balanced images. MJ simply takes bit and pieces of your instructions and grab some images from its database, while dall-e actually listens to your detailed instructions and tries to create an original image. In the long run, I think dall-e would be the better Ai art generator.
The difference is Midjourney is doing a lot more hand holding. Dalle's 3D capabilities for are exceptional if you know how to use it beyond a simple prompt.
Not only was your presentation awesome, it was super well done! My only complaint was that you favored MJ more than I did, BUT, in your conclusion, you made several valid points that I totally agreed with. Bottom line is that it is mostly subjective as it should be. I'm very, very glad that I stumbled onto you and happy to be a new subscriber. I look forward to viewing more of your videos going forward!
Why are both so terrible with creating accurate historical weapons especially muskets. They are also not too accurate on equipment, uniforms and accoutrements. Which seems strange since there are so many references on the internet, I do find that Dall-E does better with reference images than Midjourney.
AI image generation now is more of a mistake or a interpretation of many things, or more abstract like you're referring too, but make no mistake, IMO, when these technologies like image generation get so good and perfect, is when it will get boring. For me, the best images I have seen in my life have been through Midjourney, most of the elements that are in those images... are not correct, but you know what they are. It's not 3d rendered perfect intentions but those nuances make the best mistakes I've seen. NO human would ever create most of what Midjourney creates, too many influeneces, no human is that influenced. That being said, you're right, it does seem like it should be more dialed in, perhaps it's not meant too? Who knows.
As a Dall•e 3 user it’s all about you prompt, comparing the two is not really practical when they both use different ways of interpreting a prompt. You can indeed get a just as good as Mid Journey if you “know” how to prompt, using bokeh, distance etc etc I can indeed recreate just as good, and did with the cyborg image by prompting correctly. It’s just maybe Mid Journey is easier. And yes you can use certain images in Dall•e 3, like Star Wars, Marvel, and painters ‘in the style of…’
Its not fair to use the same prompt for each platform. Each platform requires specific prompt language to get what you want so using the same will not get you the best of each.
Dall-E was probably the first to make it mainstream, but it has remained very much at the level of 2022 in terms of creativity and style. While MidJourney and LeonardoAI have come very far from those early iterations. Firefly and Meta still have some catching up to do (although Firefly is exceptional at realistic / not-over-the-top beautiful portraits)
The MJ content looks, as you mentioned, more alive, more realistic, but it rly often just ignores the prompts. Cant one just use dall e and get the same effects as MJ has through further instructions, just with the quality of prompt following of dall e? MJs basic plan is useless for anything serious with the monthly limit of 200 prompts, so its like 30$+ a month, while I can get dall e for 20$ on top of chat gpt thats super useful for my IT job. But the MJ images rly look more pleasing...
I canceled my MJ for gpt pro, MJ is expensive and it has so many things you can't do, I'm not saying it's bad but Dalle is reasonably priced and it can put texts well in images as well as the flags, I always suffered with flags with MJ, it can only get first world countries flags correctly if you ask it for an African country flag it will never get it.
MJ looks more creative to me in terms of regenerating the new ideas, dallE is more illustrative and artistic looking shots for example painting and illustrations, even comic book
If you have a complicated image with multiple interacting elements -- a very specific image -- it's not even close. There are images you can make with Dall-E that are _absolutely impossible_ with MJ. Dall-E-3 is _fun_ -- MJ is more of a chore. But if you _really_ want to see those pores in the face of the busty Asian model with elf ears, or whatever, then MJ it is.
First time seeing your content, I didn’t watch past the Pixar chapter as it seemed you are very biased towards MJ even though Dalle made exactly what you asked for. Smooth rounded robot, cheerful colour scheme, well kept lawn, bright and colourful, waving etc.. MJ’s image had none of what you asked for but seemed to be your preferred image. You even hated on its childish appearance though the chapter title is Pixar.
Thanks for dropping by, I appreciate your feedback. I totally agree that DALLE follows prompts much better. As for the Pixar image, my opinion is that Pixar inputs a lot of personality into its characters and breaks away from stereo-types
Dall-E generates amazing images effortlessly, perfect for the kiddos. But don't forget the fun of Midjourney, where creativity and mistakes lead to unique artwork. So, appreciate both!
Professional Designer here. I would agree that Midjourney is ahead of Dall-e, the images are a little more mature. Dall-e via gpt is very restrictive for example I wanted a street scene with a starbucks on a corner and Dalle-3 refused to generate that. Because the starbucks logo is the IP of starbucks. MJ is much better at creating logos if you know how to prompt for that. The problem with chat GPT is it cant see the images it generates. It just assumes that it's request in the prompt was fulfilled. However it often isnt.
Why waste time or be so equivocating, mid journey is way more natural and realistic and dall-e is wholly synthetic and air brushed as if it completely misunderstands the prompt or what the real world expects. It so obvious and apparent that dall-e is woefully skewed at a fundamental level. Before it was integrated into chatGPT it was the complete opposite and was wholly realistic as per mid journey. Dall-e is unusable.
I like the depth of field in Midjourney presentation. The characters are more realistic and true to life and Dall E is more illustrative. I am a older person and seem to like what I see in Midjourney a little bit more but I do see that Dall E seems to follow instructions better than Midjourney
Great breakdown. I halted my MJ monthly for GPT Pro (Dalle). I totally see how MJ produces more interesting, realistic works of art, but with Dalle I can collaborate with a (censored) super supportive, always positive art genius. Plus, I get the worlds most powerful LLM for everything else that provides. Each has it's merit. I have to admit, I do love the output of MJ better. But the user experience (and added benefits) of Dalle take the cake.
Ya for sure, it's not perfect. And sometimes it just won't do what you want, no matter how simple. It's odd. But as they say in AI, 'this is the worst it will ever be.' So, I know in a few years this will all be ironed out for everyone. It'll just be about which package you prefer--which subscription. I wonder if these AI generators (Dalle, MJ) will in a few years time both boast text to video as well. I assume they will. @Bhodisatvas
Hmm... Pic 1 (photorealism): Win for MJ, easy. Only one of those looked anything like a picture. Pic 2 (geometric pattern): MJ made the nicer wallpaper, but failed to include circles as per the prompt. I'd still use the MJ generation every time. Pic 3 (Logos): I'd give this one to Dall-E because MJ got overly ambitious on small details. But frankly, I'd reroll both attempts. Pic 4 (Pixar): How did MJ get 'postapocalyptic' from friendly, cheerful and colorful? Dall-E win by process of elimination. Samson, did you completely forget what you prompted for while evaluating this one? Pic 5 (Cyborg): I don't like the helmet. If you're that augmented, what's the helmet for? Meanwhile, the Dall-E has no humanity remaining. That's a robot, IMO. I dislike them both. Pic 6 (Influencer): Only one of those isn't straight out of the uncanny valley. MJ win for sure. Though even MJ still needs to work on skin texture. Her ears are wonky and she looks like she buys makeup by the truckload. Pic 7 (Fantasy): I'll give this one to Dall-E for having more dragon in the image and actually following the prompt regarding forests, rivers and architecture. Even if the MJ picture is less cluttered. Pic 8 (Tattoo): Hard one. I have no idea about japanese designs, but I don't think they had scales? Also, the Dall-E one seems like it'd be a better tattoo. Pic 9 (Anthros): Neither really fit the description, I think. The MJ has the mood right, but Dall-E got the prompt right and doesn't look like a hobo. Pic 10 (surreal landscape): I prefer the MJ pic, though it doubled the double moons. Pic 11 (Gothic Cathedral): MJ Win, both in ternms of architecture and in terms of atmosphere. And I kind of quit at this point. That's 5 wins for MJ, 3 wins for Dall-E and 3 fails where I actively disliked both images.
Hi great video. Are you aware that you can invite the MJ bot to a private chatroom? That has easily eliminated the annoyance of doing work in a public chatroom.
I will say this. I write stories and need the illustrations to contain the right elements, expressions and posses. So ... although MJ images are always better looking. I choose DALL-E because accuracy of scene is more important than technically fantastic art. Very often MJ just straight up ignores my requests in favor of a beautiful face.
I'm also a novelist. I never tried DALL-E, but I like MJ for it's fine tuning features. Yes, as you say it tends to ignore my prompts at times, but there are prompting tools that can be used to fine tune whatever you want to create. For example, there are times where I get a result I really like, but with minor things that need to be fixed. Well, I can use a editing tool to fix a region that needs regenerated and take it from there.
When you spoke about pricing, you said Dall e was $20 per month with no restrictions on the amount of images, but do you not also have to pay for each image you generate, as well as the $20?
Midjourney mostly, but Dalle is better at following complex instructions. So if you absolutely want what you want, sometimes Dalle is better. I’m doing repeating patterns though much of the time, and Midjourney is the king of that, and almost the one and only. V6 seems to have upped their tiling game as well, which I’m thrilled about.
you are using over complciated polluted prompts, gpt has contaminated your promtps, that is why most of the dalle images you made were not good. Tip*Tell gpt to not alter your prompt and dont go off discussing itwith gpt, Otherwise you will end up with "GPT" looking images, insert a properly engineered prompt instead (filename is the prompt so you can check if your prompt was altered).
@@D_VAULTZ if you or anybody else is able to get artistically equal or better images out of DallE with a different promting technique, please let me know I would be very interested in this
