# MidJourney v5 Midjourney: 6 Months Later – O que mudou?
## Introdução
### Sobre o MidJourney v5 Midjourney
## Leads Orgânicos vs. Leads Pagos: Qual a diferença?
### O que são leads orgânicos?
### Como diferenciar leads orgânicos de leads pagos?
## Por que os leads orgânicos são importantes para o seu negócio?
### Benefícios dos leads orgânicos
### Como gerar mais leads orgânicos?
## Otimizando seu site para os mecanismos de busca
### Como escolher as palavras-chave corretas
### Como usar meta tags para melhorar sua classificação
### A importância da velocidade do site
## Criando conteúdo engajador
### Como escolher os tópicos certos
### Como criar títulos atraentes
### A importância das imagens e vídeos
## Utilizando as redes sociais para gerar leads orgânicos
### Como escolher as redes sociais corretas para o seu negócio
### A importância do conteúdo compartilhável
### Como medir o sucesso das suas campanhas nas redes sociais
## Construindo uma lista de e-mails
### Por que a construção de uma lista de e-mails é importante?
### Como atrair os seus clientes para a sua lista de e-mails
### Como enviar e-mails que realmente são abertos e lidos?
## Perguntas Frequentes
### O que é MidJourney v5 Midjourney?
### Como posso saber se um lead é orgânico ou pago?
### Por que os leads orgânicos são importantes?
### Como posso medir o meu sucesso na geração de leads orgânicos?
### Como posso construir uma lista de e-mails eficaz?
## Conclusão

Levar em conta as técnicas de otimização para SEO e a criação de conteúdo de qualidade pode fazer toda a diferença para impulsionar os resultados do seu negócio e gerar leads orgânicos. Com as dicas deste artigo, você pode começar a trabalhar na estratégia da sua empresa e colher resultados a longo prazo. Não perca mais tempo e comece agora mesmo a implementar essas técnicas em seu site!

br>The development pace for Midjourney has been quite remarkable! Every few weeks there’s been a new update promising new features and improvements. But how does that translate to image quality? Let’s find out!

—————————————————————————————————-
SUPPORT THE CHANNEL (& get something back in return)

3D scanned food assets:

Adobe Illustrator tutorial:

—————————————————————————————————-

My gear:

Synology 1821

Western Digital Red Plus 12TB

Audio Recorder: Zoom H8

Camera: Panasonic GH6

Headphones: Sony MDR 7506

Tablet: iPad Pro 12.9 inch

Note:
When you buy something using the affiliate links in this video or any of my other videos, I earn a small affiliate commission at no additional cost to you. Your support is appreciated!

———

Host & Creator: Dimitris Katsafouros

Website:
Twitter:

Este vídeo foi indexado através do Youtube link da fonte
MidJourney v5 ,

Midjourney,A.I.,stable diffusion ,

https://www.youtubepp.com/watch?v=d4lY2TjqNTs ,

Time flies it’s been five months already since I first started experimenting with mid-journey and since then the development Pace has been ruthless the mid Journey team just never stops it seems that every couple of weeks there’s a new update that improves the results of the algorithm which is exactly what

We as users want to see in one of my first majority videos I said I’m sure if I redo these images in a couple of months the results will probably be 10 times better so I guess now’s the time to put this statement to the test and to

Do that we’re going to use the exact same problems as before and we’ll compare the results are we going to get better looking images let’s find out Just a couple of days ago the majority developers released the fourth version of the ai’s algorithms it has a ton of improvements on pretty much every single area so in order for us to take full advantage of the ai’s latest and greatest additions we’re going to use

The V4 command at the end of The Prompt but apart from that small addition the prompts are going to be exactly the same I think the best way to start is with the very first image I shared on that initial video foreign four Images Pac-Man the AI had some

Difficulties with the placement of things the spaceship is not really hovering over the city it’s just there on the image the AI also has some trouble visualizing what that futuristic city looks like overall there’s not a lot of detail there the images have this rough looking quality but his sketches they’re great

Now let’s see how the current AI deals with the exact same prompt the difference in visual quality is quite stunning the AI managed to produce images with good Clarity great mood and a pleasing color palette but most importantly majorly clearly understands what we’re going for a huge UFO hovering

Over a city the spaceships and all the new iterations are not only better looking compared to the old versions but they’re also more clearly defined we get some good design variants and offer us to move to the next step of the process even with this first set of images it’s

Obvious how much mid-journey has improved but things get even better let’s check out the second prompt [Applause] this is the original set of images my journey is notoriously bad with hands for some reason it just cannot draw them correctly whenever they’re visible they look like a mangled mess but it’s not

Just that mid Journey has a lot of trouble rendering the other element of the prompt the old watch it also has a hard time understanding The Prompt so in the first image we’re not seeing a hand holding a watch but hand holding now let’s see what this new iteration can do

That is night and day the composition lighting and mood are absolutely wonderful my journey finally understands what an old watch looks like and there’s also no misinterpretation of the prompt the hand is holding the old watch and the chain is always visible the hands are still problematic but

They’re definitely a whole lot better than anything the older algorithm could do but it’s funny to see how easily the AI can mess up a human would never draw a hand with more than five fingers but mid Journey just keeps adding fingers like there’s no tomorrow but despite that

We’re definitely getting closer to the AI nailing the look of the real thing we’re not there yet but we’re getting there this was another problematic area for mid-journey faces were just super difficult to do and as you can see the results were quite scary the mid-journey team though quickly improved on that and

In a matter of weeks we had some huge improvements in that area but with the latest version we’re at a whole other level the eyes an area that was usually quite problematic are now rendered without any issues and basically every feature of the face is now rendered correctly but I think the

Area that deserves more praise is how the AI interprets The Prompt in the earlier iteration midrandi tried to fuse the dragon and the human into one single element but with version 4 mid journey is a lot smarter some images have dragon-like elements as we can see in

The armor of the main character and other times we have a nicely detailed dragon on display it’s really amazing to see how well mid Journey manages to identify the nuances of the vague prompt this is another area where we’ve seen some huge strides forward mid Journey

Had a tendency to spit out images that looked a bit too depressing even the happiest sounding prompts would be transformed into Moody and Bleak looking imagery to take a look at these images when I was writing the prompt I was definitely not expecting this look but

Thankfully this is now a thing of the past these are the sort of images we can get with version 4. that’s exactly what I was expecting to see a friendlier and happier color palette more playful elements and a great looking image overall some of the houses have candy elements some have

Strong colors like pink and blue and all of them have great looking designs something that was lacking in the previous set of images if we compare the new ones to the older ones it’s obvious how much mid-journey has improved there’s no comparison really I think it’s safe to say that Michonne’s

Rendering capabilities have improved tremendously the AI can now portray objects and environments in a much more convincing way let’s take as an example this prompt five months ago my journey had a huge problem visualizing this setting all of the objects were extremely messy and the actual images didn’t make much sense

Fast forward to current day and the images now are really good this scene looks like it’s in a diner the elements and the prompt look like they’re supposed to and the lighting in most of the images is spot on the rendering is not photorealistic like I would have

Expected from a prompt that was based on a photographer but still I think we’re close enough it’s a good starting point and that’s all we need from an image coming out of an AI a quick inspiration for us to start building our own image let’s now have a closer look at another

Prompt showcasing mid Journey’s new rendering capabilities this is the original set of images from our first glance it doesn’t look like there’s anything wrong with them we have some beautiful looking colors and some nice compositions but notice that the images rely on the colors to convey the magical part requested in the prompt

There’s nothing magical about the Silhouettes of the buildings now let’s see what the new algorithm came up with everything in this set of images is much more defined and like we’ve seen with the previous prompts mid journey is now better and understanding the text the magical nature of the city is portrayed

In lots of different ways we have different looking architecture we have floating lights we even have multiple planets in the sky it’s things that an actual illustrator would try to implement in the illustration in order to sell the concept it’s great to see the AI doing so well in that regard not

To mention how beautiful all these images look of course one could say that the previous more abstract images were better but keep in mind that if you’re brainstorming for an actual concept nice looking colors are not enough they’re helpful but we need more than that but

Out of curiosity I try to see if I could get something more abstract and we can definitely get there if we want to as I was working with a new version there’s one thing that might disappoint some users and that’s the ability to imitate a specific look slash artist let

Me show you what I mean here’s one of my favorite artists Simon stalinhak his work is absolutely wonderful and if you have the chance I would encourage you to go through his work it’s really really cool this is how the old version of the algorithm interpreted style and hacks work

There are some similarities mainly in the color palette but there are also several issues the robot for example is made out of houses even though there’s nothing like that in the prompt let’s now check how the new version decided to render the images here there’s no confusion with a prompt the

Robot is portrayed like an actual robot but the main problem is that the images look nothing like style and hacks work the color palette is way too bright and all elements and the images look cartoony rather than realistic I do like how some of these images look but they definitely don’t fulfill the

Main criteria of the prompt I tried this with several other artists and the results were similar here for example is how the old algorithm try to imitate Eggleston rendering and composition issues aside I do get the sense that the AI is trying to follow eggleston’s style it’s not

Doing a great job with the image but I can see what it’s trying to do now here’s how the new algorithm looks I love the overall composition colors and the fact that all objects are rendered correctly but these images look nothing like eggleston’s work not even

Remotely for me this is not a big deal what matters is that we have a general idea visualized but if this plays a bigger part in your workflow it’s good to know either way I really appreciate how good these images look the AI is finally doing a great job following the

Prompt and rendering objects correctly and that’s definitely higher on my priority list here’s an example of how these two things following the prompt and rendering images correctly can make or break an image the results of the earlier algorithm are absolutely atrocious the AI can barely render a dinosaur let alone imitate an

Art style just as a reminder this is Jeff kuhn’s work now let’s check the new algorithm the dinosaur looks like a dinosaur and we can also see the AI trying to render shinier surfaces like some objects from Coon’s work do they look like Coons no but I’m

Absolutely fine with that at least they look really really good it’s remarkable to see how much mid-journey has evolved in less than six months we went from primitive looking drawings that can barely follow the user’s prompts to beautiful renders that closely match the user’s description if midrandy’s development continues at

This pace I think we’ll see some insane stuff in a year’s time so don’t be surprised if you see me repeat this same experiment by the end of 2023 take care and I’ll see you on the next one foreign

,00:00 time flies it’s been five months already
00:03 since I first started experimenting with
00:05 mid-journey and since then the
00:07 development Pace has been ruthless the
00:10 mid Journey team just never stops it
00:13 seems that every couple of weeks there’s
00:15 a new update that improves the results
00:16 of the algorithm which is exactly what
00:19 we as users want to see in one of my
00:21 first majority videos I said I’m sure if
00:24 I redo these images in a couple of
00:26 months the results will probably be 10
00:28 times better so I guess now’s the time
00:31 to put this statement to the test and to
00:33 do that we’re going to use the exact
00:35 same problems as before and we’ll
00:37 compare the results are we going to get
00:39 better looking images let’s find out
00:41 [Music]
00:47 just a couple of days ago the majority
00:50 developers released the fourth version
00:52 of the ai’s algorithms it has a ton of
00:54 improvements on pretty much every single
00:57 area so in order for us to take full
00:58 advantage of the ai’s latest and
01:00 greatest additions we’re going to use
01:02 the V4 command at the end of The Prompt
01:05 but apart from that small addition the
01:08 prompts are going to be exactly the same
01:11 I think the best way to start is with
01:14 the very first image I shared on that
01:15 initial video
01:19 foreign
01:21 four Images Pac-Man the AI had some
01:25 difficulties with the placement of
01:27 things the spaceship is not really
01:29 hovering over the city it’s just there
01:31 on the image the AI also has some
01:34 trouble visualizing what that futuristic
01:36 city looks like overall there’s not a
01:39 lot of detail there the images have this
01:41 rough looking quality but his sketches
01:43 they’re great
01:44 now let’s see how the current AI deals
01:47 with the exact same prompt
01:51 the difference in visual quality is
01:53 quite stunning the AI managed to produce
01:56 images with good Clarity great mood and
01:59 a pleasing color palette but most
02:02 importantly majorly clearly understands
02:04 what we’re going for a huge UFO hovering
02:08 over a city the spaceships and all the
02:10 new iterations are not only better
02:12 looking compared to the old versions but
02:14 they’re also more clearly defined we get
02:16 some good design variants and offer us
02:18 to move to the next step of the process
02:21 even with this first set of images it’s
02:23 obvious how much mid-journey has
02:25 improved but things get even better
02:27 let’s check out the second prompt
02:31 [Applause]
02:34 this is the original set of images my
02:37 journey is notoriously bad with hands
02:39 for some reason it just cannot draw them
02:41 correctly whenever they’re visible they
02:44 look like a mangled mess but it’s not
02:46 just that mid Journey has a lot of
02:48 trouble rendering the other element of
02:50 the prompt the old watch it also has a
02:53 hard time understanding The Prompt so in
02:55 the first image we’re not seeing a hand
02:57 holding a watch but hand holding
03:01 now let’s see what this new iteration
03:03 can do
03:04 that is night and day the composition
03:08 lighting and mood are absolutely
03:11 wonderful my journey finally understands
03:14 what an old watch looks like and there’s
03:16 also no misinterpretation of the prompt
03:18 the hand is holding the old watch and
03:22 the chain is always visible
03:24 the hands are still problematic but
03:26 they’re definitely a whole lot better
03:27 than anything the older algorithm could
03:30 do
03:31 but it’s funny to see how easily the AI
03:34 can mess up a human would never draw a
03:37 hand with more than five fingers but mid
03:39 Journey just keeps adding fingers like
03:41 there’s no tomorrow but despite that
03:44 we’re definitely getting closer to the
03:45 AI nailing the look of the real thing
03:48 we’re not there yet but we’re getting
03:50 there
03:54 this was another problematic area for
03:56 mid-journey faces were just super
03:58 difficult to do and as you can see the
03:59 results were quite scary the mid-journey
04:02 team though quickly improved on that and
04:04 in a matter of weeks we had some huge
04:06 improvements in that area
04:07 but with the latest version we’re at a
04:10 whole other level the eyes an area that
04:12 was usually quite problematic are now
04:15 rendered without any issues and
04:17 basically every feature of the face is
04:20 now rendered correctly but I think the
04:22 area that deserves more praise is how
04:24 the AI interprets The Prompt in the
04:27 earlier iteration midrandi tried to fuse
04:29 the dragon and the human into one single
04:32 element but with version 4 mid journey
04:35 is a lot smarter some images have
04:37 dragon-like elements as we can see in
04:40 the armor of the main character and
04:42 other times we have a nicely detailed
04:44 dragon on display it’s really amazing to
04:47 see how well mid Journey manages to
04:49 identify the nuances of the vague prompt
04:57 this is another area where we’ve seen
04:59 some huge strides forward mid Journey
05:02 had a tendency to spit out images that
05:03 looked a bit too depressing even the
05:06 happiest sounding prompts would be
05:08 transformed into Moody and Bleak looking
05:10 imagery to take a look at these images
05:13 when I was writing the prompt I was
05:16 definitely not expecting this look but
05:19 thankfully this is now a thing of the
05:21 past these are the sort of images we can
05:24 get with version 4.
05:26 that’s exactly what I was expecting to
05:29 see a friendlier and happier color
05:31 palette more playful elements and a
05:34 great looking image overall some of the
05:36 houses have candy elements some have
05:38 strong colors like pink and blue and all
05:41 of them have great looking designs
05:43 something that was lacking in the
05:45 previous set of images if we compare the
05:48 new ones to the older ones it’s obvious
05:50 how much mid-journey has improved
05:52 there’s no comparison really
05:54 I think it’s safe to say that Michonne’s
05:57 rendering capabilities have improved
05:59 tremendously the AI can now portray
06:02 objects and environments in a much more
06:04 convincing way let’s take as an example
06:07 this prompt
06:13 five months ago my journey had a huge
06:15 problem visualizing this setting all of
06:18 the objects were extremely messy and the
06:20 actual images didn’t make much sense
06:23 fast forward to current day and the
06:26 images now are really good this scene
06:28 looks like it’s in a diner the elements
06:31 and the prompt look like they’re
06:32 supposed to and the lighting in most of
06:35 the images is spot on the rendering is
06:37 not photorealistic like I would have
06:39 expected from a prompt that was based on
06:41 a photographer but still I think we’re
06:43 close enough it’s a good starting point
06:45 and that’s all we need from an image
06:47 coming out of an AI a quick inspiration
06:49 for us to start building our own image
06:52 let’s now have a closer look at another
06:55 prompt showcasing mid Journey’s new
06:57 rendering capabilities
07:02 this is the original set of images from
07:04 our first glance it doesn’t look like
07:06 there’s anything wrong with them we have
07:08 some beautiful looking colors and some
07:10 nice compositions but notice that the
07:12 images rely on the colors to convey the
07:15 magical part requested in the prompt
07:17 there’s nothing magical about the
07:18 Silhouettes of the buildings
07:21 now let’s see what the new algorithm
07:23 came up with
07:24 everything in this set of images is much
07:27 more defined and like we’ve seen with
07:29 the previous prompts mid journey is now
07:31 better and understanding the text the
07:34 magical nature of the city is portrayed
07:36 in lots of different ways we have
07:38 different looking architecture we have
07:40 floating lights we even have multiple
07:43 planets in the sky it’s things that an
07:45 actual illustrator would try to
07:47 implement in the illustration in order
07:49 to sell the concept it’s great to see
07:51 the AI doing so well in that regard not
07:54 to mention how beautiful all these
07:56 images look
07:57 of course one could say that the
07:59 previous more abstract images were
08:01 better but keep in mind that if you’re
08:03 brainstorming for an actual concept nice
08:05 looking colors are not enough they’re
08:07 helpful but we need more than that but
08:10 out of curiosity I try to see if I could
08:12 get something more abstract
08:15 and we can definitely get there if we
08:17 want to
08:19 as I was working with a new version
08:21 there’s one thing that might disappoint
08:23 some users and that’s the ability to
08:25 imitate a specific look slash artist let
08:28 me show you what I mean here’s one of my
08:30 favorite artists Simon stalinhak his
08:33 work is absolutely wonderful and if you
08:35 have the chance I would encourage you to
08:36 go through his work it’s really really
08:39 cool
08:44 this is how the old version of the
08:46 algorithm interpreted style and hacks
08:48 work
08:49 there are some similarities mainly in
08:52 the color palette but there are also
08:53 several issues the robot for example is
08:56 made out of houses even though there’s
08:58 nothing like that in the prompt
09:01 let’s now check how the new version
09:03 decided to render the images here
09:05 there’s no confusion with a prompt the
09:07 robot is portrayed like an actual robot
09:10 but the main problem is that the images
09:12 look nothing like style and hacks work
09:15 the color palette is way too bright and
09:18 all elements and the images look
09:20 cartoony rather than realistic
09:22 I do like how some of these images look
09:25 but they definitely don’t fulfill the
09:27 main criteria of the prompt
09:29 I tried this with several other artists
09:32 and the results were similar
09:39 here for example is how the old
09:40 algorithm try to imitate Eggleston
09:43 rendering and composition issues aside I
09:46 do get the sense that the AI is trying
09:48 to follow eggleston’s style it’s not
09:51 doing a great job with the image but I
09:53 can see what it’s trying to do now
09:55 here’s how the new algorithm looks
09:58 I love the overall composition colors
10:01 and the fact that all objects are
10:03 rendered correctly but these images look
10:05 nothing like eggleston’s work not even
10:08 remotely for me this is not a big deal
10:10 what matters is that we have a general
10:12 idea visualized but if this plays a
10:15 bigger part in your workflow it’s good
10:17 to know either way I really appreciate
10:20 how good these images look the AI is
10:24 finally doing a great job following the
10:26 prompt and rendering objects correctly
10:29 and that’s definitely higher on my
10:32 priority list
10:33 here’s an example of how these two
10:36 things following the prompt and
10:38 rendering images correctly can make or
10:40 break an image
10:45 the results of the earlier algorithm are
10:48 absolutely atrocious the AI can barely
10:51 render a dinosaur let alone imitate an
10:54 art style just as a reminder this is
10:57 Jeff kuhn’s work
10:59 now let’s check the new algorithm the
11:02 dinosaur looks like a dinosaur and we
11:04 can also see the AI trying to render
11:06 shinier surfaces like some objects from
11:09 Coon’s work
11:10 do they look like Coons no but I’m
11:13 absolutely fine with that at least they
11:16 look really really good
11:20 it’s remarkable to see how much
11:22 mid-journey has evolved in less than six
11:25 months we went from primitive looking
11:26 drawings that can barely follow the
11:28 user’s prompts to beautiful renders that
11:31 closely match the user’s description
11:33 if midrandy’s development continues at
11:35 this pace I think we’ll see some insane
11:37 stuff in a year’s time so don’t be
11:40 surprised if you see me repeat this same
11:41 experiment by the end of 2023 take care
11:45 and I’ll see you on the next one
11:49 foreign
11:52 [Music]
, , , #Midjourney #months , [agora]