bro just use stable diffusion. it is veryyyy flexible and if you are a pro, you can make high quality images like midjourney. and you can make everything. and it is FREE
Thank you!! I am doing mnemonics, and havent’ had to use blood yet, but How about a toilet, and diarrhoea? (I’m trying to do a demented/confused (can’t make that happen) pelican with diarrhoea - with a red rash around his neck (can’t seem to make that happen either) with diarrhoea (won’t allow it, so I write ‘brown blobs on the floor - which sort of works) for a medical condition called pellagra. I have been able to say a white pelican with a rash on wings, in a white tiled room with a shower and brown blobs on the floor, which is working ok, but its not the best tbh. Im wanting Pixar style, So I might try Flux and see what I can make!!!! 🎉
I am using mimicpc which can also achieve such effect. You can try it for free. In comparison, I think the use process of mimicpc is more streamlined and friendly.
Yes, I think flux is awesome, I tried Stable diffusion on Mimicpc, and of course this product also includes popular tools for AI such as RVC, Fooocus, and others. I think it handles detail quite well too, I can't get away from detailing images in my profession and this fulfills exactly what I need for my career.
*I dunno...Runway gave Luma the #1 spot after they release an underwhelming Gen 3....The keyframe function in Luma is a game changer. Great video tho and thanks for all the info! Gave me some ideas 🤔*
Midjourney new updates are taking the image generator platforms to a whole new level. 6.2 Has a startling percentage of spectacular images per prompt. For now it's mid journey, and everyone else are just fighting for 2nd place.
@@curiousrefuge Absolutely. But also realistic dialogue. Unless I'm missing something, and I watch all your vids and some of the other gurus, it just aint even close right now.
@@giorgiteliashvili5268 Should be many tutorials on UA-cam for a full install by now. I'd recommend the Flux Dev Q8 version since it works best for GPU with 12GB VRAM or less while still being close to the full uncompressed version. ComfyUI is the frontend that most use to run this but you can also check out Forge too.
The shoe images were generated with blue tint because the shots were outside. Outside light other than and sunrise and sunset is pretty blue more than what people realize. It's why pro cameras to do it right you have to tell the camera what is white.
@@curiousrefuge It can be in interesting excercise ... if you close your eyes and imagine what white looks like in your inner photoshop and then look at something that you know is white objectively ... often enough you will notice that "white" is either more blue or yellow depending on the light source.
Thanks for the info. Also... 33:28 => You said "the future of pitching", which also applies to the video game world (for you aspiring game designers out there). You can mock up a game HUD in Photopea and then run it through the video generators to create "in game footage" of your proposed title.
I’m grateful for the time and effort you've put into guiding me. The skills I've learned from you will surely be very helpful in the future........following you .Alll . i had done 7 videos....to try...........so great......Ai bro!!!!!!!!
An intriguing analysis of Midjourney! Another great option is Stable Diffusion, which is understood for its flexibility and high-quality outputs. #Midjourney #StableDiffusion
The first schnell version you some how over looked the hand had 6 fingers. For me inpainting is critical .... at least until they come up with a way to make characters consistent. I am doing image to video mostly these days.
Great video!.. I tried this prompt on the Flux1 free version: A sketch of three elderly men named Stan, one running to the left, one running to the right, and one running forward facing the viewer. All Stans are 60 years old, with two locks of hair on their heads, wearing long-sleeve shirts and slacks. The background is a minimalistic, soft pencil shading with motion lines emphasizing their movement, hd quality, dynamic composition, exaggerated facial expressions, quick pencil strokes, cartoonish style, capturing the essence of an active elderly person .............. unfortunalty, same problem that I was facing with playground and Dall-e and other free text to image softwares. The problem is: the 3 men are either all running in 1 direction or 1 is running to one direction and the other 2 are running forward... I have not tried this in midjourney yet because I am hesitant to buy it and then find out there is a tool that can do what I want it to do.. please advise... thanks
I counted 31 "basicallies" in this 32 minute video, I wonder if one per minute counts as world record (I also tried to count the adverbs that were not needed but I stopped at 50, I basically got tired)
"They (flux) raised $31 million dollars", just a reminder, Stable Diffusion raised $101 million in 2022 and was the darling of UA-camrs, and they raised other $80 million in 2024 before taking a nosedive. Now every UA-camr is flexing Flux as the next AI, hopefully history doesn't repeat itself, I like to see Flux take AI to the next level.
I tried Flux (schnell and def) with ComfyUI locally. Can do batches of images just fine. Your are correct with size limits, 2048x2048 works but I get a old CRT mask look that kinda breaks the image details.
MESHY isthe most outstanding tool from all this news, IT REALLY IS !!!!! i usedto spend weeks and months modeling stupid details and stuff, NOW ITS ONE CLICK and done!!!! how good is that !
For a German speaker it is very painful to hear schnell (means „fast“) with hard l at the end. It has a soft consonant L at the end and L reads like L in word leaf. I hope it helps)
i agree, midjourney has too much goin on in the backend to make oneclick wonder app and that makes it not as PRO as they claim i would put it in the semi-pro, even tho results are good, but no technicality aporach to it makesit SEMI-PRO
Although Midjourney is an effective AI picture generator, SmythOS provides more capabilities and creative options. It resembles having a personal helper for design.
2:09 Awkward battle shovel? That girl has 6 fingers (at least) on each hand, she'll kill you with that. In the future, that's what battle-axes might look like. Small surface, for ease of handling in confined spaces.
Does FLUX do any art styles from centuries of art history like Midjourney does? All I've seen from FLUX besides photorealism, is the same old anime and Pixar like stuff. It seems that's all FLUX was trained on, maybe? With no sref or cref, it doesn't seem like FLUX would support much beyond photographic work.
Flux is decent but not midjourney. I tried netanyahu in hitler style with a demon behind him (common theme these days) and to its credit it didn't refuse but it's often a little muppet caricature, at times cartoon style even though I stated photo. At best it resembled satanyahu. It struggled to do a hitler stache, it tended to be a mexican stache. I guess many of these generators don't aim to get likeness very good but they should.
At 1:55 the sci-fi woman has six digits on her right hand ( the thumb is obscured ). I thought this problem was limited to AI videos. Disappointing on an AI still image at this stage. I'm surprised the Host didn't pick that up. Another great show, though.
Everything I saw here was terrible. The problem with A.I. generation is that it's nothing new. Everything is copied from an artist's work. And that virtual production app that you showed was so terrible and built just to hype this stuff. Earlier, we didn't know how this technology worked, so it was kinda intriguing, but I saw all the AI trailers, and they looked like copied footage from hollywood films. I just hope courts restrict apps like midjourney if they are not compensating artists as these generators are stealing their work.
If you want to use blood in Midjourney just put in your prompt, "red syrup dripping", your welcome. Works in Runway too.
I had the same luck with Runway. 😂 Yes! It's not blocked!
Like in movies XD
bro just use stable diffusion. it is veryyyy flexible and if you are a pro, you can make high quality images like midjourney. and you can make everything. and it is FREE
And if you want the blood to look funny, just put in your prompt, "Quentin Tarantino."
Thank you!! I am doing mnemonics, and havent’ had to use blood yet, but How about a toilet, and diarrhoea? (I’m trying to do a demented/confused (can’t make that happen) pelican with diarrhoea - with a red rash around his neck (can’t seem to make that happen either) with diarrhoea (won’t allow it, so I write ‘brown blobs on the floor - which sort of works) for a medical condition called pellagra. I have been able to say a white pelican with a rash on wings, in a white tiled room with a shower and brown blobs on the floor, which is working ok, but its not the best tbh. Im wanting Pixar style, So I might try Flux and see what I can make!!!! 🎉
I've been using Midjourney for a long time, and so far I like it more than all other photo generation services.
We love MJ!
Flux is a whole different kettle of fish. With the flexibility of ComfyUI, theres no comparison.
I am using mimicpc which can also achieve such effect. You can try it for free. In comparison, I think the use process of mimicpc is more streamlined and friendly.
Great video! Hopefully we can choose first AND last frame in Runway soon.
We can't wait for that too!
I love it - a battle shovel! I really appreciate your caveats about Flux.1. I've come to really trust your advice.
Brilliant updates, thank you sir for this valuable info!
amazing comparison, flux is really good and im excited for the future versions.
So are we :)
Yes, I think flux is awesome, I tried Stable diffusion on Mimicpc, and of course this product also includes popular tools for AI such as RVC, Fooocus, and others. I think it handles detail quite well too, I can't get away from detailing images in my profession and this fulfills exactly what I need for my career.
*I dunno...Runway gave Luma the #1 spot after they release an underwhelming Gen 3....The keyframe function in Luma is a game changer. Great video tho and thanks for all the info! Gave me some ideas 🤔*
Midjourney new updates are taking the image generator platforms to a whole new level. 6.2 Has a startling percentage of spectacular images per prompt. For now it's mid journey, and everyone else are just fighting for 2nd place.
I agree
The new textures are staggeringly impressive,
We believe "control" is a huge factor for big professional projects.
@@curiousrefuge Absolutely. But also realistic dialogue. Unless I'm missing something, and I watch all your vids and some of the other gurus, it just aint even close right now.
Both the Scnell and Dev versions are open source. Also the Dev version is the recommended open source version and not just for developers.
Totally understood - we tried to use the most accessible and easy version to test that doesn't require any technical skills.
that comment was a help. could you hint where to get Dev for pc install?
@@giorgiteliashvili5268 Should be many tutorials on UA-cam for a full install by now. I'd recommend the Flux Dev Q8 version since it works best for GPU with 12GB VRAM or less while still being close to the full uncompressed version. ComfyUI is the frontend that most use to run this but you can also check out Forge too.
The shoe images were generated with blue tint because the shots were outside. Outside light other than and sunrise and sunset is pretty blue more than what people realize. It's why pro cameras to do it right you have to tell the camera what is white.
Great feedback!
@@curiousrefuge It can be in interesting excercise ... if you close your eyes and imagine what white looks like in your inner photoshop and then look at something that you know is white objectively ... often enough you will notice that "white" is either more blue or yellow depending on the light source.
I use grok 2, and it's really good too, and it won't censorship copyright characters and blood, not like chatGPT.
We like Grok!
Very informative! Thank you for your time.
I was very surprised by the difference in quality between Flux and Midjourney for the woman in a yellow shirt with "Curious Refuge"
Glad you enjoyed that!
Another banger Matt 👌👍
Thank you! My name is Caleb, but I totally give Matt haha
Thanks for the info. Also...
33:28 => You said "the future of pitching", which also applies to the video game world (for you aspiring game designers out there). You can mock up a game HUD in Photopea and then run it through the video generators to create "in game footage" of your proposed title.
Great point!
@@curiousrefuge It just so happens to be my new favorite thing to do.
First time seeing your reviews, they are really really good. The companies should pay you, it's pretty much work 😆
We prefer not to have sponsors so that we ensure our content is always as unbiased as possible! :)
The head of the LA women with a yellow shirt at 8.27is ridiculously out of proportion.
8:27
Agreed!
I’m grateful for the time and effort you've put into guiding me. The skills I've learned from you will surely be very helpful in the future........following you .Alll . i had done 7 videos....to try...........so great......Ai bro!!!!!!!!
You are so welcome
NICE BIDEO Thank you my friend
Our pleasure!
Great work caleb
Thanks!
An intriguing analysis of Midjourney! Another great option is Stable Diffusion, which is understood for its flexibility and high-quality outputs. #Midjourney #StableDiffusion
We'll have to do a better rundown of it sometime!
Nice comparison vid!
Thanks for watching!
🎯 Key points for quick navigation:
00:00 *New AI competitor*
00:13 *Flux image generator*
00:26 *Backed by investors*
01:21 *High-quality generations*
02:02 *Three Flux versions*
02:55 *Less content filtering*
03:35 *Customizable prompts*
07:04 *Good prompt adherence*
11:12 *Midjourney comparison*
13:48 *Aspect ratio flexibility*
16:37 *Typography evaluation*
18:54 *Limitations noted*
19:46 *Future potential discussed*
20:00 *Expensive image generation*
22:00 *Runway new features*
23:24 *Incredible shot details*
23:38 *Video jitter noticed*
24:18 *Celebrity voice project*
25:10 *Custom data dashboards*
25:38 *Improved animation quality*
26:07 *Virtual production tool*
27:31 *AI relighting functionality*
29:00 *3D model generation*
30:25 *Turbo modes emerging*
31:34 *AI matting improvements*
32:16 *Innovative film concepts*
33:12 *Audience engagement increase*
34:06 *Dynamic camera movements*
Made with HARPA AI
Thanks for the breakdown!
The first schnell version you some how over looked the hand had 6 fingers.
For me inpainting is critical .... at least until they come up with a way to make characters consistent. I am doing image to video mostly these days.
True - the six fingers is a big no no!
@@curiousrefuge With the exception if we are talking about a character in the Princess Bride of course
You doing an amazing job I wish your video was translated
What language? We are working on it :)
Schnell is evolving into Chanel ?
😊
Haha it’s the luxury version.
@@curiousrefuge Fun fact: "Schnell" means "fast" in German
Nexus 1945 sounds like someone wanted to do Strike Witches, but without the middle school girl soldiers.
Interesting!
Great video!.. I tried this prompt on the Flux1 free version: A sketch of three elderly men named Stan, one running to the left, one running to the right, and one running forward facing the viewer. All Stans are 60 years old, with two locks of hair on their heads, wearing long-sleeve shirts and slacks. The background is a minimalistic, soft pencil shading with motion lines emphasizing their movement, hd quality, dynamic composition, exaggerated facial expressions, quick pencil strokes, cartoonish style, capturing the essence of an active elderly person .............. unfortunalty, same problem that I was facing with playground and Dall-e and other free text to image softwares. The problem is: the 3 men are either all running in 1 direction or 1 is running to one direction and the other 2 are running forward... I have not tried this in midjourney yet because I am hesitant to buy it and then find out there is a tool that can do what I want it to do.. please advise... thanks
Get on our discord and our community can help!
@@curiousrefuge what is your discord link? I don't see it in the description box. thanks
I counted 31 "basicallies" in this 32 minute video, I wonder if one per minute counts as world record (I also tried to count the adverbs that were not needed but I stopped at 50, I basically got tired)
You counted? We appreciate the effort!
"They (flux) raised $31 million dollars", just a reminder, Stable Diffusion raised $101 million in 2022 and was the darling of UA-camrs, and they raised other $80 million in 2024 before taking a nosedive. Now every UA-camr is flexing Flux as the next AI, hopefully history doesn't repeat itself, I like to see Flux take AI to the next level.
Very interesting point!
I tried Flux (schnell and def) with ComfyUI locally. Can do batches of images just fine.
Your are correct with size limits, 2048x2048 works but I get a old CRT mask look that kinda breaks the image details.
thanks for testing and letting us know!
MESHY isthe most outstanding tool from all this news, IT REALLY IS !!!!! i usedto spend weeks and months modeling stupid details and stuff, NOW ITS ONE CLICK and done!!!! how good is that !
We appreciate you commenting!
Nice content👌👌✔✔
Thanks ✌️
For a German speaker it is very painful to hear schnell (means „fast“) with hard l at the end. It has a soft consonant L at the end and L reads like L in word leaf. I hope it helps)
We appreciate the tip, we'll try and adjust how we say it :)
I think must of us prefer the schnell or dev option because they better 🤷🏼 and totally free open source
True! For testing though this was a far easier method :)
It will be too cold when image generators are compatible with professional screenwriting software. Screenwriting scripts are super specific.
One day!
@@curiousrefuge Soon!
i agree, midjourney has too much goin on in the backend to make oneclick wonder app and that makes it not as PRO as they claim i would put it in the semi-pro, even tho results are good, but no technicality aporach to it makesit SEMI-PRO
Gotta love MJ :)
What's better between Mid journey and Leonardo?
We typically use MJ but both are great. You can try both out!
Although Midjourney is an effective AI picture generator, SmythOS provides more capabilities and creative options. It resembles having a personal helper for design.
We'll have to check it out!
Can you please include links to the ai created films of the week? I would like to watch "Legend of the odd mirror" and cannot find it.🙂
Thanks, we'll start adding them :). You can also find them posted on our AI gallery on our website :)
@@curiousrefugecool, thank you!! 👍🏼🙂🍀🧡
subbed :)
Thanks for the sub!
@@curiousrefuge my pleasure
2:09 Awkward battle shovel? That girl has 6 fingers (at least) on each hand, she'll kill you with that. In the future, that's what battle-axes might look like. Small surface, for ease of handling in confined spaces.
Certainly still has a way to go!
Anything that is going to charge my credit card to try out is a deal breaker.
Totally understood
Grok 2.0: Nope!
We like Grok too :)
@@curiousrefuge 👍
Weird Science is coming
Very soon :)
Does FLUX do any art styles from centuries of art history like Midjourney does? All I've seen from FLUX besides photorealism, is the same old anime and Pixar like stuff. It seems that's all FLUX was trained on, maybe? With no sref or cref, it doesn't seem like FLUX would support much beyond photographic work.
It certainly doesn't have the control that MJ has. But we can try more testing to see how it does with non photorealism.
MJ still can generate consistent characters without kind of workaround work flow.
We still love MJ!
Flux is decent but not midjourney. I tried netanyahu in hitler style with a demon behind him (common theme these days) and to its credit it didn't refuse but it's often a little muppet caricature, at times cartoon style even though I stated photo. At best it resembled satanyahu. It struggled to do a hitler stache, it tended to be a mexican stache. I guess many of these generators don't aim to get likeness very good but they should.
Interesting test - we'll stay out of politics though :)
Midjourney cant generate the vegetable ‘Lady Finger’ through prompts.. Anyone who is able to do so should reply to me here?
They are all capable of good and bad generations :)
At 1:55 the sci-fi woman has six digits on her right hand ( the thumb is obscured ). I thought this problem was limited to AI videos. Disappointing on an AI still image at this stage. I'm surprised the Host didn't pick that up. Another great show, though.
Definitely still issues across all AI generators. Gotta keep rolling :)
The only difference that matters: one is free, the other isnt.
Control is very important to us though!
a NICKEL per image lmao..
It's a lot!
i love you 🥰
Thanks for watching, we love you too! :)
I watched this whole video to learn MJ is better when you already knew it was. 😡
Thanks for watching!
we used Flux Pro to create this Images and used Kling to generate the video: ua-cam.com/video/E42SSnxTdxE/v-deo.html
Very cool!
Obiously not!!!
That's why we tested - to each their own :)
you put bundereds of link except the flux website. I give a dislike for this reason
Sorry!
Everything I saw here was terrible. The problem with A.I. generation is that it's nothing new. Everything is copied from an artist's work. And that virtual production app that you showed was so terrible and built just to hype this stuff. Earlier, we didn't know how this technology worked, so it was kinda intriguing, but I saw all the AI trailers, and they looked like copied footage from hollywood films. I just hope courts restrict apps like midjourney if they are not compensating artists as these generators are stealing their work.
We understand your perspective but respectfully disagree.
Great work caleb
Great work caleb