Great video, watched every second of it. I agree on most points but although AI Art is not my main focus I would dare to predict that this is going to cause the biggest decline in MJ subscribers we have seen. The artistic value of Dall-E 3 can be on par with Midjourney but the coherence, text generation, access to popular IPs and ease of use will make it very hard for most people to maintain 2 subscriptions instead of just 1. On the flip side it should be mentioned that for creative professionals midjourney obviously remains the best tool for extremely artistic and detailed results + Dall-E 3 generations do NOT come with a commercial license. PS: Love the AI Video B-Roll
Yes, this is definitely going to make a huge dent in MJ's earnings because the vast majority of people barely notice the subtleties of MJ's aesthetics and DALL-E 3 is fully sufficient for them. To be honest, I can see myself downgrading too, simply because this will change my workflow. I'll be using both in combination quite a bit.
I'm using Dall-E 3 in Bing & best part is it is free. I really like it so far, crazy how much control we can have over image creation. Game changer imo.
You are absolutely right- it's hard to judge Dalle3 just from advertising material. They made a lot of noise with this announcement yet "normal folks" still can't access it. Why?
After your course, it's not hard to spot the differences between Dall-E3 and Midjourney. I think Dall-E 3 is moving in the right direction & for the average ai art buff, it's gonna be a real game changer. And yet, once you learn how its gears work, I think you can still get way more bang for the buck out of Midjourney. Thanks for these videos, Chris. Please keep them coming.
At the start of the text-to-image explosion, I tried several. Both Dall-E and Stable Diffusion were far too complicated for me and I almost gave up as I was not impressed with others I tried. I found Midjourney and quickly started creating the kind of things that pleased me. In just over a year, I feel that I've become quite adept at using MJ and even forged a new and different career direction. I for one will be staying faithful to MJ unless it changes drastically and not for the better. Thanks for the vid :)
As far as prompting it is way better than any other image prompter out there. Anyone can get access even now. Each sentence is read as its own prompt in the entire prompt. They still have many issues too kick mid journey. Like only a 500 character prompt limit in DALL-E 3, No adjustments to an existing image, and it is only one ratio. This may change when the roll out of the main product comes out but for now it is only good for testing.
Too strictly, but decent generates close to prompts. Anyone having an issue with it or cooldown on users after a couple of images generated. "You can't submit anymore prompt "
You're the first to complain about the audio in over half a year. Especially since it's put through an AI audio enhancer. But I'll pass it on to my editor.
i think Dalle3 in the future is going to be more friendly user (like canva) and users with 0 graphic knowledge just type a prompt to get an expectable result, they even might mix chat gpt and dalle. also most of the Employees are promgramers mijourney in the other hand is more like photoshop and people with more graphic knowledge use it and it gives more art style results. founder of midjourney spend lots of time on just text to image and that is why they are #1 in this category. their main focus is more good result rather than a well coded program (they dont even use their own platform for midjourny) overall midjourney is all in image
I've always found the term "Prompt Engineer" totally overblown, elitist w@nkery. You don't need to be a technician to create great images in programs like MJ. The term will become redundant in very short order as the language models adapt to common language commands.
It was always a doomed profession in my opinion. If these systems become more aligned and they will … this alignment includes following and anticipating user intent… hence the prompt engineering was an in between necessity for little and early models that needed help to actually “understand”.
Love your Contant, hate your hard cuts. You should be getting better at your production and not having to have editing fix your mistakes. Hopefully you don't do that in your paid content.
I'll pass it on to my editor. I don't do the editing. But if there's one thing I've learned over the past 10 months, it's that there's always something that bothers someone. No matter what. Initially people complained about music. So I removed it. Now people complain that there's no music. 🤷🏻♂️ So, I'm not really sure what you expect from me. I don't have a huge budget or team to create these videos. And there's no point in comparing myself to Johnny Harris and other YT rockstars. My primary objective is to share knowledge, not to become the next VICE with the production quality of Netflix. RU-vid isn't even the primary thing I do. It's more like #3 on the list. As for my paid content, the satisfaction of my students speaks for itself. So I have absolutely no qualms.
It's prompting is phenomenal but it still has a long way to go to compete with mid journey. I think its biggest issue is the character count limit, most good detailed prompts are hard to explain. Like the old phrase says a picture is worth 1000 words if it cant hold that many words at least it will be difficult to get your images the way you want them entirely. Another issue I noticed is that it does not follow hex code coloring which limits the colors a user can associate with an object.
I think the threat depends on what you’re making. Dall-E blew me away at launch. A chair that looks like a banana. Perfect. A pig riding a motorcycle. Perfect. But I started using Midjourney… paradoxically for both control and creativity. If you need a coherent stock image fast, Dall-E is amazing. But if you want to experiment, Midjourney is better.
Midjourney has less restrictions, but coherence and text makes dall-e a winner. Dall-e is cheaper and it can creates entire comic book panels with dialogues. That's amazing and now I'm going to unsubscribe Midjourney.
Aesthetics aside, DALL-E 3's capabilities in terms of prompt coherence are on another level. MJ can't compete in that respect at all right now. I'll make that very clear next Saturday.
DALL-E 3ish (Bing Image Creator, free for now but only square images) excels with multiple objects, and detailed instructions, in a single image . I use Midjourney daily, but it's frustrating to generate multiple objects images and it ignores much of the details requested. Midjourney needs to improve NLP, multiple objects & correct text verbatim soon, otherwise DALL-E 3 will race ahead.
Can you tell me what I'm doing wrong? I can get Bing Image Creator to actually produce images at all. It feels broken. Endless waiting time, constant popups. I've tested multiple browsers. So far, it has generated only a single set of images for me. It's super frustrating.
@@TokenizedAI I'm probably way late to the party of this comment, but their servers were massively overloaded shortly after announcement. I think it's calmed down a bit now.
I've tried the prompt in dall-e 3 (at 05:10 in your video ) and it seems to give the same kind of result : "At the corner stall, a young woman with fiery red hair, dressed in a signature velvet cloak, is haggling with the grumpy old vendor. The grumpy old vendor, a tall, sophisticated man, is wearing a sharp suit sport a noteworthy moustache and is animatedly conversing on his steampunk telephone. The sidewalks bustling with pedestrians enjoying the nightlife. A bustling city street uner the shine of a full moon."
As a retired Creative Director, I absolutely agree with every point you've made. 10:16 is the only one I would say made the entire video a little binary as to what the 'best' would be, considering they are currently targeting different demographics and target markets, albeit with an overlap. 'Best' would be measured on the use case, which you've touched upon earlier in the video. I also think that there will be a point where people might end up using different platforms for separate workflows for the same 'project'. _edit: You did mention that none of us know right after that, so I'll leave my moot comment about that up here anyway!_ Great video! 💯 Thank you. 👍
I do not think it will be because of the censorship. If you are not able to fully express what you have inn mind, but abide by a filter of ''safe content'' people will eventually go for a less restricted tool.