Test Driving Ideogram v1.0 - The New King of Coherence?

igram has just turned one they've just gone from their pre-release versions of version 0.1 and 0.2 to version one which means now they're ready to go out into the world and show EV what they're capable they're no longer in Alpha they're no longer in pre-release everything's gone huge and the question is it worth your hard earn Dollar in this video we're going to test the free version and really see if it's worth your time and then we're going to talk about the price of upgrading and if it's worth doing when it launched igram was really known as the first to market with the ability to create images with text inside of them they did it before Dolly 3 they did it before mid Journey version 6 but was it enough to stay ahead now they've added a whole bunch of features claiming they have the best prompt coherence which means if you have multiple pieces to the image they go in the right place can it win we're going to look at improving past images I've created some of their promises testing the new magic prompt feature and comparing it to my image Wars tests from a couple of days ago to see if it can compete with mid Journey Dolly 3 stable diffusion and of course Leonardo so we're going to do a whole bunch of really cool tests in this videos that we're going to find out if idiogram is worth your time let's hit that music cuz this is going to be a good [Music] one so very exciting just a few hours ago idiogram announced version 1.0 this is a big deal because they means they moved out of Alpha and beta where you give a partial number they were 0.1 then 0.2 for quite a few months now and to jump to 1.0 means they finally feel ready to show the world what they have and what they could do and they've made this cool little video showing all the things they can do demonstrating basically the text coherence is what they're really excited by which means that this text here in the demonstration not only are all the words spelled correctly it looks like the same font which is one of the most important things to look at they have some other examples that they're showing this video which is worth watching they posted this to X or maybe you call it Twitter and they have this blog post announcing the itagram 1.0 just in time for leap day they announced on February 28th which is awesome you can see the demonstrations from their video help people become more creative in a lot of different styles videogram is known for images that have text inside them they were first to Market with this feature shortly afterwards Dolly came out with the feature about after they released to Market and then four or five months later mid Journey add the feature but they were the first ones and really dominated for a little while the question is have they recaptured their space they make a lot of promises about the character error rate how they have the lowest error rate it's not zero they just say it's lower than everyone else their examples are always going to be the best version of everything that they do they're also talking about two other really important features idiogram has always been one trick pony it's always been seen certainly by me I'm guilty of this of seeing is something that you only use if you want text and the image honestly only if I wanted to have a cartoonish or Pixar looks have things changed here they have a look real to me this is quite real they're doing some very interesting things they're talking about Artistic Styles which I really like I really like what's happening in this artistic style especially very interesting with the puzzle and what's really exciting is that they give you a bunch of examples of their best and they give you the prompt you can test this prompt in indogram to see how well it works to see if they're telling the truth they also have a new feature called Magic prompt which means you can write a small prompt and or write a longer one this is similar to what mid Journey does mid Journey calls it raw mid journey is natural state is to be a beautification mode or try to make the image better than what you described raw is where you turn that off and it actually listens to the prompt they give a name for when you turn it off magic prompt is the name for when you turn it on with idiogram let's take a look inside of idiogram we're looking for great images and one of the best features the thing that separates idram for everyone else as these little hearts I can save any heart as a seed image that I'll use later you're going to see a lot of this image because this is from their blog post people are testing it we can look and if you see something you like you hit the heart you can use it later I have a large collection like this is quite good this is very much the style I expect from them whereas this is not so these are things happening in different areas you can actually build up your own version of the mid Journey visual dictionary which there's a link to right below this video which is my collection of mid Journey promps here you actually stored in Ingram and I do this quite a lot I mostly use it for storing text prompts so anything that has text in it that I really like and to get better instead of looking at explore I can go to top which are things other people have liked and say is there an image that's really good you can see them using the example prompt from the blog post there same one here a lot of these are from the blog post examples so it's this image but we're looking for our images generated with version one only images in the last eight hours are made in the newer version version 1.0 version 1.0 these are quite good this is very good because the O is actually like an on button that's why it's got the line through the top of it I hit the heart and I can use that later to generate a prompt I'm also looking for other styles of images cuz now they're claiming to be more than one trick pony I'm building my own personal collection perfect text generation is possible seeing is believing try .0 for yourself people are really excited by what's coming out and what they're playing around with and they're testing because the made some really big promises but can they live up to it this is the first and most important feature and if we click on generate you can see what's changed you can now do magic prompt on offer Auto which is where it goes all your prompt is good enough we don't need to run it or we're going to help you you have three ratios and you can choose which model 0.1 Point 2 1. know visibility private or public you can only go private if you have a paid plan let's go and look at my profile you can see 25 of 25 prompts that means I can generate 25 batches of four or 100 images per day which is very generous this is a really great tool you can test the only thing you can't really do is private generation so if you make an image you want it to be a secret then you have to pay to do that and let's look at pricing they have three tiers free basic and plus to get private generation you have to pay $20 a month if you want to get access to the ADR editor that's interesting that's a new tool you have to be on a paid plan to upload images you have to be on the paid plan at the highest level and at this level then you get priority generation and Standard Generation this means faster and relaxed mode on Mid journey I have the same thing when I use up my 15 hours so mid Journey does it by time versus by number of images I have hit the limit before because we're building the mid Journey visual dictionary and when I go over that I just drop to relax mode instead of taking 30 seconds making image takes about a minute and a half here you can see unlimited so if you're really going to go crazy it's quite worth it as soon as you go paid you can download a high quality image prior access new features improvements and there are no restrictions very interesting what's the better plan I think the most important feature that might appeal to you if you're thinking about going between 8 or $20 is the private generation and image upload I'll have to know more about this to know if it's worth it because and this is something I want to play around with in test if it can use my image as a seed you know that in many of my videos I've shown most AI image generators make me ugly I don't want to pay a whole bunch of money to find out that it sees me as ugly so I will test this on the monthly plan if you're thinking about going for the higher level start out the $20 a month before you jump to annual if these features don't work that well then it's not worth it I would definitely be on the monthly plan to test them and then possibly accelerate I'm going to test the idiogram editor in a later video but I want to show you guys the free version first but we're going to start with is my past image Generations you can see all the things I've done in the past I'm actually going to just rerun old prompts and you can see all of my old prompts here let me see if I can make it a little better the very first test I made was making t-shirts for happy dot day I can open this up and see the exact prompt I used that's the dimensions and I can rerun the same prompt so I written it right here you can see they have these buttons to add features to it and you can see the type phography was clicked for this one because that's what that does because I'm copying the old prompt I don't have to do that and I can just set it to which version I want obviously I want 1.0 you can see a whole bunch more choices when you go to version 1.0 happy. day with a poka dot party typography let's hit it this is what's called Standard Generation speed so it's going to be slower but already it's going quite fast which is interesting because I know that a lot of people are using the tool right now they brought a bunch of users in with their new announcement so it's probably the busiest it's going to be for a while and you can see that the jumping quality is basically insane I could throw this on a T-shirt and sell this right easily the polka dots are very good the font is pretty close to perfect the a is a little bit wonky but it's juny it doesn't look broken it looks intentional so it falls within that range these other images are close this y turned into an X and it pulled in party we're going to go back and look for some other ones and you can see a lot of the images have that Pixar or airbrush color feel this is another project I was working on this is a blog post where I talk about creating the perfect logo we're going to see if we can make a better logo now because this is really important and again we're in 1.1 we're in version 1.0 and we're going to click generate to see if we can make this and if it understands that Argentinian blue is a specific heximal color it's only counted as one prompt so even version 1.0 it's not counting it as a higher number like 1.0 costs three promps or five proms the first thing we're looking at is coherence so serve no Baker this has the correct words serve no Baker no spelling problems here you have serve and then this is gobbley serve Baker is fine and then here it pulled in words outside the quote like the word Vector got pulled in so what we're seeing is some very different things this is the best of the four it's not life Chang in it's definitely okay we're going to look at some other old prompts so here's dot day here's one that I really enjoyed I like this one a lot this is again idiogram version one we're seeing the difference between version 0.1 and version one going to do the exact same generation I'm very impressed with how fast it's generating that I notice a huge difference we're looking for a retro 1970s Apollo 13 a motorcycle style design with the text to the moon this is very good it says Apollo 10 instead of a 13 or maybe that's supposed to be a H with an extra L if you take the two of the moon off the top this is a great image this is unbelievable as far as the quality down here the text is not perfect we're seeing definitely a jump in quality this was trying to create a design for one of my programs called cyber staffing agency and we're going to go through the same process and hit generate and we're just going to let it run the same thing happens with a lot of image generation tools at 99% it looks terrible then at 100% it looks great here the text is really solid here the text is off this is pretty close to perfect and this is very solid as well the two apps are touching each other but otherwise this is great I like this one the best this is one I would use so this is something that I would just use as a logo it's great so I'm really pleased with that so here's a t-shirt design it says retro Vector t-shirt logo design with dragon with glasses black background sun in the background of the image with the text Year of the Dragon graffiti photo poster 3D those are the three buttons that I clicked when I made it for is t-shirt design that I can immediately throw on a shirt without needing to do any editing very close with this design but you can see there's two G's and dragon which is quite annoying and then this Dragon looks a little bit wonky in the text what I normally do is generate multiple versions of an image to get the result I want and you'll notice that down here is that I have many versions when I'm working through a process like the 99 prompts logo this is the exact one that I use for 99 prompts this is for cre version two just a couple of weeks ago because I'm working on the release of 9N prompts version two let's see how much of a difference just a few weeks can make and we're in version 1.0 one by one click generate one of the differences between idiogram and mid journey is concurrent I can have up to 10 images Genera at the same time in mid Journey or in the queue while four are going at the same time this is a huge jump in quality now I'm seeing the stuff I wanted to see this is so much better and it may have to do with the fact that my prompts were better for this version but these are all really good logos this looks amazing this looks amazing this is the only one that's not amazing and that's because there's some issues with the background image just making the text a little hard to see but there's no spelling issues all four of these have the right spelling very cool I was working on redoing the logo for my LinkedIn newsletter and this is metallic blue modern Sleek logo with a text 100k from chat GPT hypography 25 images a day seems generous until you want to try all the features it starts to feel like not enough all right so this one is good this is really good for this image I would remove the background and put something else behind it but it's not terrible it's just a matter of I would tweak it and I just want to always be honest with you guys of how I'm going to use it this is the logo for the artificial intelligence podcast I paid someone to redo this I'm going to take this exact image rerun it in version 1.0 I'm actually going to use this as a seed image seeing this I realized that I didn't need to hire the person on Fiverr to redraw the image I can now do it in here the best of these four is number two it's not perfect but I can regenerate this five or 10 times and I'll get the result I want you either need to make one logo per day or be on a paid plan just depends the amount of work you want to run through this tool this is an image I really liked I really like where this is going it reminds me of the the animation style in the movie trolls you can see I need to turn off that image it generate this is really a test to see is it better for prompts I've already run through it is it better than the past already we're seeing that the spelling is correct for all of these the first one may have two wise after the crazy now what I'm going to do is remix the image and hit gener to see if using the seed image how much of a difference that makes there just a huge jumping quality really good we've used 10 out of our image Generations I don't want to use them all let's go to next test we're going to do next is test all the images from my recent video called image Wars where I was testing prompts from Dolly 3 stable Cascade mid Journey Leonardo and we're just going to test how ADR holds up against the competition this is the first test it was a floral little acraft with spring flow I'm copying and pasting in the prompt exactly as it was in that test so you can see how it Compares here's the slide from last week where I thought mid Journey was the best one Leonardo was a little bit off in the shape of the A and now we're going to look here and I actually think this is the best image it's not what I asked for exactly they want a little creative the issue with this image is that the background is too dark it's hard to see it this is letter four but we're in the same range our next test is the avocado test an illustration of an avocado sitting in therapist's chair saying I feel so empty inside and a pit size hold at Center therapist of spoon scribbles notes we have multiple characters in the screen as well as next this is very complicated prompt none of these got it exactly right all four of these have the pit still inside not the hole where the pit is supposed to be even the dolly 3 image which is the only one that includes the therapist with the spoon doesn't have the text and this is an example from the do from the dolly page so this one should be the best adog just won so it's won this test from the image generation Wars I would say this is better coherence it has the missing pit there's some misinterpretation of the spoon you can see the spoon is here here the spoon is coming inside the person's or like going in into eating out the tummy but this is very close we went it has a therapist instead of a spoon therapist but overall we're seeing the empty pit three out of four times the eye feel so empty inside is really good here very interesting to see how it's interpreting this so this is very solid very surprising I never know what's going to happen I'm running these tests in real time to give you authenticity this prompt is Post Punk core beautiful cute Reggae Woman with black long hair dreadlocks and Boots lying in a hammock diesel Punk flower Punk of these Dolly 3 is the one that had the flower Punk involved and the diesel Punk the rest of these just look like photographs of a woman in Hammock which is not bad okay I have to zoom in a little bit to see in this case this is very solid this is pretty much exactly what the prompt was looking for although it looks like she's not wearing shorts she's missing a leg goes here and then disappears don't like that image and I don't really like this one you can see that perhaps diesel Punk and flower punk are outside of how idiogram is trained so it's not quite perfect I would say it's not the winner of that test our next test was very simple Race sign Street sty logo Minecraft style picks on our original example my two favorites were mid journey and Leonardo AI mid Journey has a really cool pixel style image this is not quite Minecraft this is a little bit more higher resolution than Minecraft it looks pretty cool and these are very 8bit this really didn't get that pixel style idea let's zoom in and see if there's any element of that pulling in so definitely not this image no now these are awesome images but I'm testing is the coherence it doesn't have the pixel style or pixel art style it didn't win that prompt and that's why we do these tests next test is for this paper stop art let's look at the example again this is a very long prompt multidimensional car cell paper CH illustration Sunshine background two hands paper cutting cell hands are passing a heart hands of Love midow Vector painting Thomas concade the key words here Thomas concade is a very strong important word almost the rest of the words are supplementary we have two hands we have a heart element in the middle in all four of these this starts to come down to personal taste in my opinion Dolly 3 is quite good but there's a problem with the fingers and then otherwise Leonardo is the winner out of the four this is the only one that's close to a paper art style we're seeing that when we go way outside the area of expertise for the tool we're not getting a great result you might have to use magic prompt which will be the last test we run this is a really difficult prompt because there's a lot of elements it's a really a big test of coherence or knowing where to put things when a rain DED world she takes a moment it's a post-apocalyptic world she has her left hand holding the umbrella and her right hand has a mask that looks somewhere between a piglet and a dog and then she's got a purple windbreaker that's supposed to go to Mid thighs and she's Barefoot with an empty stair you can see she's Barefoot here the feet aren't showing she has the mask in all four Images which is astounding the umbrella unfortunately is in the wrong hand but that's very easy to fix by flipping the image the mask does look halfway between a pig and a dog this is not quite a dog this is not quite a pig this throws things off because she has two the other thing is that the mask is supposed to be above her head which it accomplishes in three out of the images the Cod is to the right Point overall for this prompt which is one of the hardest ones in my testing set ADR is the winner now let's go to the next test looking at our past example we have Dolly 3 stable Cascade Le on Mid Journey mid Journey is quite good St Leonardo again one I think it's very good at this style of kirami where it's a bunch of piece of construction paper layered on top of each other so it's like a 3D image like my son's diarama which sometimes you can see over my shoulder behind me a bunch of foxes that's what this reminds me of this is not kirigami so it fails on the area of coherence but this is a cool image I believe a lot of this has to do with magic prompt possibly affecting the image and this image it almost looks like like the same woman in all four this just looks like an older version of the same woman you can see the nose shape is the same for these four the glasses are same for these two almost the same for her really feels like the same person so I wonder if ITR is going to use the same woman not quite this is much more colorful much more unique this doesn't feel like a photograph it feels like a drawing is it the winner not really this is too scary I want to get away from these because they're too scary going to see our next test got me into trouble with Dolly 3 it shouted at me twice it wouldn't make the image it got mad because I had the words Marvel in there Iron Man these don't look like Iron Man they were just an inspiration but it triggered that word it was also triggered by the word Gucci pattern on the armor none of these was a winner for me none of this was an image that I would actually use but here we have it so this feels like Gucci Samurai armor I can tell right away the this prompt would need a lot of refining because it doesn't have as much of the Iron Man pull this just looks like Gucci Samurai armor but it's giving me a lot to work with this prompt needs some refinement which is okay and we're going to go to the logo test this is a test for this channel because I was looking for an AI news logo to put in the top corner of the thumbnails when it's a news report each of these is okay none of these is usable right away each of these I would have to play around with to make the logo exactly what I want this is crazy this is very cool but I think it's too hard to read let me know what you guys think in the comments below which of these logos should I switch to this is very cool I think it's too cool and then the rest of these it seems like a painting in the background so the logos are all boring there's something triggering this issue with the prompt because I made a lot of logos in the past and I think that it's the way I have the commas round logo with the text AI news not even going to put a color in I'm going to turn magic prompt on and we're going to see what happens it's going to to rewrite the prompt before creating the image so we can actually see what it changed the prompt to when I click on one of these images see now it says a capting vibrant 3D render of round logo featuring the text AI news a unique combination of anime graffi and Yu-Gi-Oh Styles look how long the prompt has become I'm going to add some elements to it blue neon use magic prompt on again and you're going to see a massive jump as well and we're going to see how easy it is to prompt engineer with idiogram so we you don't have to add really complicated description just adding a word or two should make a significant difference and the magic will take it the rest of the way so these are very cool this is very good it's a little hard on the readability scale almost looked like all news this could be interpreted as an L however this is so cool this is very tempting for me to use I want to go to my profile and look at what I've liked I want to test this one I'm just going to hit remix for this image because because this is a different style notice that it's not that Pixar style it's very cool just going to hit remix and very important is you can choose the image weight on a scale of 1 to five how important is the original image right now it's a three out of five I'm going to leave it there we have magic prompt on let's let it just go absolutely crazy and see how close it can get we merge these two things woo these are looking really cool so the quality jumped up of the image it's really personal taste which of these four you like but all four of these have great text great usability these would be very easy to use as a t-shirt overall very impressed by this image let's try this one this is another type of image that's a great scene image because you can see the text is pretty good there's a double P here and this says perfect it's perfect but the structure itself I think is good and I'm going to regenerate this again the problem in the image is pulling into this so the reason this went wrong is that the spelling issues pulled over from our older generated image and that's important to understand is that if you use an old image that has a spelling problem in it that spelling problem may pull in even though the prompt is correctly smelled now we know let's go back to the homepage I believe that their organization here is very good for finding inspiration what is happening in the feed you can see where people are getting images that are quite realistic this looks like a real person so the ability to pull in this is a very good Style image here you can see disneyanime style I'm going to save this because I may use it I look for that and to see how can people create it now this is a paper image this is very interesting because they effectively created a new style I definitely need to grab this they successfully activated that kirami style that we were looking for earlier now I know how to do that this is how you can become a really good prompt engineer I think this organization is better than any of the other platforms for this one particular ability which is finding what other people have done and building on top of it and this is how I design my prompts inside of idiogram I like this platform I'm really interested in this let me know what you guys think do you want me to do more testing do you want me to grab the paid version and do a deep dive into what their editor looks like and what happens when I try and beauy my own face then that will be in part two one of the biggest challenges that I face with my job is testing all of these tools and I'm always on the fence I want to buy them but then I think is this a tool that I'm going to use enough that it justifies the purch because I have so many tools I use in my Arsenal that I spend most of my time testing new tools I don't have time to spend really just using one tool anymore as much as I want to I would love to spend just three months mastering idiogram and just getting perfect at it but I don't have the bandwidth is this worth the investment is it worth the upgrade that's the most important question for me compared to what you get for M Journey this is very solid it does something slightly different it's a really good tool to consider adding to your cool kit 100% the free version should be part of your tool kit it's been part of mine since the day it launched way back in August and I believe that for sure 25 images a day you can see I did a lot of tests and I did it run out I still have a couple left to use after this video because I still want to play around with some things for myself I love it the question is about the paid version I'm going to have to upgrade and do a test I'm going to do the monthly version at $20 a month and see what it does with the images of myself and that'll be in part two of this video but I want to first test the free version the question is has it gotten better yes for sure I believe right now this is the best free image generator when we're comparing free to free when we're comparing this to the Leonardo free image generation when we're comparing this to stable Cascade which is a free open source image generator inogram is just really good I'm very surprised but very pleased that they haven't taken away the free tier like mid Journey did at a certain point overall this is a really great tool is it perfect no we saw that if you use a past image that has a misspelling that past misspelling will pull into the new image unfortunately I can't test what happens when I upload images myself so I have to wait for the next video on this topic but golly this is a very interesting tool but the most important thing is what do you guys think image generation is very subjective it's do you like the image I've always really liked what idiogram can do but I just felt that Pixar kind of cartoonish style was too strong I can see now they're moving away from that they're really claiming to have great f realism and I want to see if you guys see that in the images and the demonstrations and the test I showed you did this tool impress you or were you thinking it's okay but it's not lifechanging I know there are some videos out there that say this is the best thing since slic bread it's not really my approach this is not a sales video this is a test where I'm making a decision myself do I think it's worth upgrading am I going to use this as my main logo generator right now most of the logos that I use are generated using Bing image generator even though I have Dolly 3 the paid version I find that being image generator makes better logos than the other tools however I believe that idiogram with version one can push ahead of that but it requires a little bit more finesse on my part requires a little bit of better prompting a little bit more skill from me but I believe that when I apply more skill it will become a better image and Logo generator so that's really going to become part of my workflow very interesting to see how things are changing and I want to just take that perspective overall I'm very pleased with this tool I think the free version needs to be added to Your Arsenal if you haven't created an account do it now just in case they take away the free accounts at some point you want to lock that in so your grandfather I love this tool I'm really excited to see what it does for a while I thought they were in trouble when they started raising more money I was thinking they only have one feature they don't have a product because other tools can add that ability which other tools have done mid Journey has it now Dolly has it now and I'm sure that Leonardo will continue to add and improve in that area as it adds more features and more Generations very important to see that this tool is taking things seriously they use the massive investment I think in an effective way there's a very good chance this tool takes a dominant position because it's really focusing on some very interesting things and for a lot of people the most important element is coherence which means if I say a dog on top of the table horse below the table it creates an image that has those two things in the right place a lot of tests I've seen of that a lot of their examples that's there I encourage you to run those tests yourself so that you do more than just watch this video you actually try this tool it's absolutely free I encourage you to sign up check it out let me know your thoughts below this video if it's worth your time if the juice is worth the squeeze or if you looked at the images and go I hate these all of that really matters because this is the future of graphic design and the question is if this tool is going to be part of that feature let me know in the comments I do read every comment I want to see what you guys have to say thank you so much for staying all the way to the end of this extended video I appreciate everything you guys do I'll see you in the next video thank you so much for watching if you like that video I think you're going to like this one or maybe even this one check them out

Channel: ServeNoMaster

Views: 741

Keywords:

Id: 2Y9vA5wEQ3k

Length: 30min 9sec (1809 seconds)

Published: Fri Mar 01 2024