Could you please return 10 for each of these prompts, I give you my best, ones that should get out of it interesting vividness:
1) Bright macro shot of a plush toy robot pikachu eating a hamburger in a nurse outfit against a white brick wall with mud splashed on pikachu from a tire on the road. 8K HD incredibly detailed.
2) Macro shot of the cool pikachu wearing black chains and laughing as seen in a truck selfie in the desert next to a sand castle with piranha plants seen through the heat. 8K incredibly detailed.
3) Future 2377 hospital with beds in glass containers, white spheres that hold tools, robot maids, cameras everywhere, and blue scrubbing systems moving around the walls. Lots of detail and systems.
4) A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. HD, detailed.
5) Microscopic water bear meets a bacteria that looks like pikachu next to a bacteria hospital pouring out different colored creatures with spikes, furr, etc. HD detailed.
6) Big black furry monsters with tall heads and white patches dripping with water and seaweed are picking up a dumpster in an alley. Realistic HD photo.
7) 3) A motor connects to a hydraulics pump, which connects to blue energy rods soaking in pink liquid. It’s smoking. Macro, detailed.
8) Video game case rated E, grey rimmed, macro shot, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo.
9) Video game case rated M, dark red rimmed, macro shot, a glossy black world that endlessly goes back into the distance with many black temples, gates, and chests. HD photo.
10) A plate with fries, nuggets, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph.
“Video game case rated M, dark red rimmed, macro shot, a glossy black world that endlessly goes back into the distance with many black temples, gates, and chests. HD photo.”
“A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. HD, detailed.” It’s not super coping with all the details – could maybe do better with more repetition in the prompt? – but it’s got the vibe.
I am going to register an advance prediction that many of these contain way too many details (both in terms of number of objects requested, and in terms of specific relationships between said objects) and are going to overwhelm the poor image model. I’ll run them as-is, but I might also try modified/simplified versions if I think I can get something more in the spirit of your requests that way.
“Big black furry monsters with tall heads and white patches dripping with water and seaweed are picking up a dumpster in an alley. Realistic HD photo.”
“Microscopic water bear meets a bacteria that looks like pikachu next to a bacteria hospital pouring out different colored creatures with spikes, furr, etc. HD detailed.”
“Future 2377 hospital with beds in glass containers, white spheres that hold tools, robot maids, cameras everywhere, and blue scrubbing systems moving around the walls. Lots of detail and systems.”
With some changes to the prompt, “A cool goth pikachu wearing black chains and laughing, sitting in a truck in the desert, next to a heat-shimmery sand castle with piranha plants. 8K incredibly detailed.”
It tends to depict Pichu rather than Pikachu. But I note that Pichu’s electric attacks damage itself, at least in Super Smash Bros (and I find a quote from the Bulbapedia article saying “it cannot discharge without being shocked itself”), which caused a friend to refer to Pichu as “emo Pikachu”. Perhaps “goth Pikachu” ended up referring to the same thing...
“Bright macro shot of a plush toy robot pikachu eating a hamburger in a nurse outfit against a white brick wall with mud splashed on pikachu from a tire on the road. 8K HD incredibly detailed.” Yeah—DALL-E seems to be landing at best a handful of the details you wanted, and in some of these it seems to be returning something almost random!
“A plush toy robot pikachu wearing a mud-splashed nurse outfit and eating a hamburger, against a white brick backdrop. Detailed HD footage.” It’s done much better here! I’m not sure any of the images managed the “mud-splashed” bit, but they’ve all got a reasonable Pikachu-robot, plus the hamburger and the white brick wall, and some of them are managing the nurse outfit.
Could you do the other prompts in my post, I want to push the model, maybe you missed them due to comment collapse. Or if want me to pick only a few let me know. This is so cool.
Wow! The glossy black temple one, wow! This is beyond belief, impossible! It not only came close to what I had imagined but forget the lava, it’s better! Just what I want. Looks like a hard game. The others are also very impressive, I Love the dumpster one it came very good, and the hospital one, and the food one is just grand. The 2 video game case ones, good but not good haha I meant not those cases, how did your brain interpret the outputs—you saw it was wrong right (lol) ?. Here is a few more and let’s try to get one of those games made right this time. Also I’m adding onto the food one something interesting and attempting to elongate that good one:
1) A video game sitting against a wall, rated E, grey rimmed, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo, macro.
2) A plate with fries, nuggets, gold fork and knife, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph. Pikachu is leaning into the plate eating the food.
3) A gold room full of red rubies, gold coins, white crystals, silver spiked chests, and gold toilets lined up.
4) A floating liquid metal blob in a laboratory is 3D printing cameras, memory, and sensors. Scientists are trying to guide it. HD, detailed.
5) Arial view over a world consisting of glossy black temples, thunder, round purple chambers, spiked lava rivers, and flat paths that maze around and monsters guarding gates. HD, detailed.
6) Inside the bright gold temple restaurant is gold tables, crystal walls, waterfalls coming out the walls, robot maids, and lots of fries and red ruby decorations. HD, detailed.
7) Glossy black dragon statue shooting red laser beams from its eyes into a glossy black wall, making it crack open exposing a gold vault. In the rain at night, HD, detailed.
8) A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. Big black bosses wearing gold chains and crowns are walking into the temple. Raining, HD, detailed.
Aaaaaaand final one. (I would kind of prefer if you keep any future requests to one or two prompts.)
“A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. Big black bosses wearing gold chains and crowns are walking into the temple. Raining, HD, detailed.”
Black robots wearing gold chains and red robes sitting in thrones made of white crystal with gold spikes lined up. The robots are holding plates with fries and ice cream over white sinks in front of their thrones facing a mirror, in a red luxury bathroom full of gold coins and doors, and white and red ruby pots.
Also can you do two Variations below showing all 10 results? (Note: I super-resolutioned one, so if you have the full version saved, check which is more detailed truly):
Plus your other request, “Black robots wearing gold chains and red robes sitting in thrones made of white crystal with gold spikes lined up. The robots are holding plates with fries and ice cream over white sinks in front of their thrones facing a mirror, in a red luxury bathroom full of gold coins and doors, and white and red ruby pots.” Honestly pretty impressed with the level of detail in the image!
Ok. That’s good no text prompt was used. I wonder what would happen if you now tried the gold room image again with it’s text prompt below, maybe it would guide the 10 Variations better? Though it seems as if you have, the Variations show toilets even though there is none in the input image, why is that? Here was the prompt, please try it (or without if you think you included text):
‘A gold room full of red rubies, gold coins, white crystals, silver spiked chests, and gold toilets lined up.’
“Black robots wearing gold chains and red robes sitting in thrones made of white crystal with gold spikes lined up. The robots are holding plates with fries and ice cream over white sinks in front of their thrones facing a mirror, in a red luxury bathroom full of gold coins and doors, and white and red ruby pots.”
“Inside the bright gold temple restaurant is gold tables, crystal walls, waterfalls coming out the walls, robot maids, and lots of fries and red ruby decorations. HD, detailed.”
“A gold room full of red rubies, gold coins, white crystals, silver spiked chests, and gold toilets lined up.” I think it’s confused on the color scheme—the room itself doesn’t appear to be gold in any of these.
“A plate with fries, nuggets, gold fork and knife, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph. Pikachu is leaning into the plate eating the food.” I think this is closer to what you were envisioning? Though, uh, mildly horrifying in a few, and also one of them made Pikachu a rubber duck?
Minor edit because ‘shooting’ appears to be a banned keyword: “Glossy black dragon statue flinging red laser beams from its eyes into a glossy black wall, making it crack open exposing a gold vault. In the rain at night, HD, detailed.”
“Arial view over a world consisting of glossy black temples, thunder, round purple chambers, spiked lava rivers, and flat paths that maze around and monsters guarding gates. HD, detailed.”
I modified #4 a bit to try to hint harder, since the initial round mostly gave me only the liquid blobs. It’s still struggling with the details, especially at including any scientists, so I think there are too many weird/not-usually-combined elements here for it to manage without much more skilled and careful prompting.
“There is a floating liquid metal blob in a laboratory. The floating liquid metal blob is is 3D printing cameras, memory, and sensors. There are scientists in the laboratory trying to guide the metal blob. HD, detailed.”
This is a lot of requests and I’m at work, so I’ll run them over the next few hours. (Honestly I’m not a video games person and had no idea that “case” was the same thing as...rating? and also I have no idea what an E rating is, I don’t recognize that one from movies.)
“A video game sitting against a wall, rated E, grey rimmed, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo, macro.” I don’t think it super knows what you want here…
A “case” in this context is the plastic clamshell that holds/protects the disc when not in use (DALL-E thinks this instead means some sort of container found within a video game environment).
The E rating (for “Everyone”) is similar to the G rating for movies.
I hypothesise that the more details a prompt contains, the more likely DALL-E is to throw a wobbly and produce something almost totally random. But honestly, I’m very impressed with the outcome of most of these prompts. The picachu eating a hamburger is the only one of the above that really “failed”, and a couple of the outputs picked up about half the requested details.
Could you please return 10 for each of these prompts, I give you my best, ones that should get out of it interesting vividness:
1) Bright macro shot of a plush toy robot pikachu eating a hamburger in a nurse outfit against a white brick wall with mud splashed on pikachu from a tire on the road. 8K HD incredibly detailed.
2) Macro shot of the cool pikachu wearing black chains and laughing as seen in a truck selfie in the desert next to a sand castle with piranha plants seen through the heat. 8K incredibly detailed.
3) Future 2377 hospital with beds in glass containers, white spheres that hold tools, robot maids, cameras everywhere, and blue scrubbing systems moving around the walls. Lots of detail and systems.
4) A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. HD, detailed.
5) Microscopic water bear meets a bacteria that looks like pikachu next to a bacteria hospital pouring out different colored creatures with spikes, furr, etc. HD detailed.
6) Big black furry monsters with tall heads and white patches dripping with water and seaweed are picking up a dumpster in an alley. Realistic HD photo.
7) 3) A motor connects to a hydraulics pump, which connects to blue energy rods soaking in pink liquid. It’s smoking. Macro, detailed.
8) Video game case rated E, grey rimmed, macro shot, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo.
9) Video game case rated M, dark red rimmed, macro shot, a glossy black world that endlessly goes back into the distance with many black temples, gates, and chests. HD photo.
10) A plate with fries, nuggets, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph.
“Video game case rated M, dark red rimmed, macro shot, a glossy black world that endlessly goes back into the distance with many black temples, gates, and chests. HD photo.”
“A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. HD, detailed.” It’s not super coping with all the details – could maybe do better with more repetition in the prompt? – but it’s got the vibe.
I am going to register an advance prediction that many of these contain way too many details (both in terms of number of objects requested, and in terms of specific relationships between said objects) and are going to overwhelm the poor image model. I’ll run them as-is, but I might also try modified/simplified versions if I think I can get something more in the spirit of your requests that way.
“A plate with fries, nuggets, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph.”
“Video game case rated E, grey rimmed, macro shot, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo.”
“A motor connects to a hydraulics pump, which connects to blue energy rods soaking in pink liquid. It’s smoking. Macro, detailed.”
“Big black furry monsters with tall heads and white patches dripping with water and seaweed are picking up a dumpster in an alley. Realistic HD photo.”
“Microscopic water bear meets a bacteria that looks like pikachu next to a bacteria hospital pouring out different colored creatures with spikes, furr, etc. HD detailed.”
“Future 2377 hospital with beds in glass containers, white spheres that hold tools, robot maids, cameras everywhere, and blue scrubbing systems moving around the walls. Lots of detail and systems.”
I think this is my favourite: https://labs.openai.com/s/KQfiNLLHurkhwSW7Cj38GWA8
With some changes to the prompt, “A cool goth pikachu wearing black chains and laughing, sitting in a truck in the desert, next to a heat-shimmery sand castle with piranha plants. 8K incredibly detailed.”
It tends to depict Pichu rather than Pikachu. But I note that Pichu’s electric attacks damage itself, at least in Super Smash Bros (and I find a quote from the Bulbapedia article saying “it cannot discharge without being shocked itself”), which caused a friend to refer to Pichu as “emo Pikachu”. Perhaps “goth Pikachu” ended up referring to the same thing...
“Bright macro shot of a plush toy robot pikachu eating a hamburger in a nurse outfit against a white brick wall with mud splashed on pikachu from a tire on the road. 8K HD incredibly detailed.” Yeah—DALL-E seems to be landing at best a handful of the details you wanted, and in some of these it seems to be returning something almost random!
“A plush toy robot pikachu wearing a mud-splashed nurse outfit and eating a hamburger, against a white brick backdrop. Detailed HD footage.” It’s done much better here! I’m not sure any of the images managed the “mud-splashed” bit, but they’ve all got a reasonable Pikachu-robot, plus the hamburger and the white brick wall, and some of them are managing the nurse outfit.
Could you do the other prompts in my post, I want to push the model, maybe you missed them due to comment collapse. Or if want me to pick only a few let me know. This is so cool.
I’ll come back to them! There’s just a whole lot of comments on this post to process.
Wow! The glossy black temple one, wow! This is beyond belief, impossible! It not only came close to what I had imagined but forget the lava, it’s better! Just what I want. Looks like a hard game. The others are also very impressive, I Love the dumpster one it came very good, and the hospital one, and the food one is just grand. The 2 video game case ones, good but not good haha I meant not those cases, how did your brain interpret the outputs—you saw it was wrong right (lol) ?. Here is a few more and let’s try to get one of those games made right this time. Also I’m adding onto the food one something interesting and attempting to elongate that good one:
1) A video game sitting against a wall, rated E, grey rimmed, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo, macro.
2) A plate with fries, nuggets, gold fork and knife, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph. Pikachu is leaning into the plate eating the food.
3) A gold room full of red rubies, gold coins, white crystals, silver spiked chests, and gold toilets lined up.
4) A floating liquid metal blob in a laboratory is 3D printing cameras, memory, and sensors. Scientists are trying to guide it. HD, detailed.
5) Arial view over a world consisting of glossy black temples, thunder, round purple chambers, spiked lava rivers, and flat paths that maze around and monsters guarding gates. HD, detailed.
6) Inside the bright gold temple restaurant is gold tables, crystal walls, waterfalls coming out the walls, robot maids, and lots of fries and red ruby decorations. HD, detailed.
7) Glossy black dragon statue shooting red laser beams from its eyes into a glossy black wall, making it crack open exposing a gold vault. In the rain at night, HD, detailed.
8) A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. Big black bosses wearing gold chains and crowns are walking into the temple. Raining, HD, detailed.
Aaaaaaand final one. (I would kind of prefer if you keep any future requests to one or two prompts.)
“A glossy black temple surrounded by lava and thunder with silver spiked chests on the ground next to the gate. Big black bosses wearing gold chains and crowns are walking into the temple. Raining, HD, detailed.”
Ok! One last one to document its limits further:
Black robots wearing gold chains and red robes sitting in thrones made of white crystal with gold spikes lined up. The robots are holding plates with fries and ice cream over white sinks in front of their thrones facing a mirror, in a red luxury bathroom full of gold coins and doors, and white and red ruby pots.
Also can you do two Variations below showing all 10 results? (Note: I super-resolutioned one, so if you have the full version saved, check which is more detailed truly):
https://ibb.co/jVFpQP8
https://ibb.co/Tb36ZQw
(uploaded using imgBB)
Plus your other request, “Black robots wearing gold chains and red robes sitting in thrones made of white crystal with gold spikes lined up. The robots are holding plates with fries and ice cream over white sinks in front of their thrones facing a mirror, in a red luxury bathroom full of gold coins and doors, and white and red ruby pots.” Honestly pretty impressed with the level of detail in the image!
For the second request, I’m not sure I follow—are these results from previous prompt rounds that I ran?
The gold room one, yes please, and the other is a mario game that would be interesting to see if it can make Variations of too. (show all 10)
Gotcha! Gold room variations here:
And the Mario game variations:
Was a text prompt used along side the image input to make these Variations? Or just image input? Very interesting results BTW.
just the image—I had uploaded them as new images bc it cleared my session and I didn’t have the originals anymore.
Ok. That’s good no text prompt was used. I wonder what would happen if you now tried the gold room image again with it’s text prompt below, maybe it would guide the 10 Variations better? Though it seems as if you have, the Variations show toilets even though there is none in the input image, why is that? Here was the prompt, please try it (or without if you think you included text):
‘A gold room full of red rubies, gold coins, white crystals, silver spiked chests, and gold toilets lined up.’
The gold room one, yes please, and the other is a mario game that would be interesting to see if it can make Variations of too. (show all 10)
“Black robots wearing gold chains and red robes sitting in thrones made of white crystal with gold spikes lined up. The robots are holding plates with fries and ice cream over white sinks in front of their thrones facing a mirror, in a red luxury bathroom full of gold coins and doors, and white and red ruby pots.”
“Inside the bright gold temple restaurant is gold tables, crystal walls, waterfalls coming out the walls, robot maids, and lots of fries and red ruby decorations. HD, detailed.”
“A gold room full of red rubies, gold coins, white crystals, silver spiked chests, and gold toilets lined up.” I think it’s confused on the color scheme—the room itself doesn’t appear to be gold in any of these.
“A plate with fries, nuggets, gold fork and knife, steak and pikachu-shaped cake covered in ketchup and salt topped with ice cream. Close-up photograph. Pikachu is leaning into the plate eating the food.” I think this is closer to what you were envisioning? Though, uh, mildly horrifying in a few, and also one of them made Pikachu a rubber duck?
Minor edit because ‘shooting’ appears to be a banned keyword: “Glossy black dragon statue flinging red laser beams from its eyes into a glossy black wall, making it crack open exposing a gold vault. In the rain at night, HD, detailed.”
“Arial view over a world consisting of glossy black temples, thunder, round purple chambers, spiked lava rivers, and flat paths that maze around and monsters guarding gates. HD, detailed.”
I modified #4 a bit to try to hint harder, since the initial round mostly gave me only the liquid blobs. It’s still struggling with the details, especially at including any scientists, so I think there are too many weird/not-usually-combined elements here for it to manage without much more skilled and careful prompting.
“There is a floating liquid metal blob in a laboratory. The floating liquid metal blob is is 3D printing cameras, memory, and sensors. There are scientists in the laboratory trying to guide the metal blob. HD, detailed.”
This is a lot of requests and I’m at work, so I’ll run them over the next few hours. (Honestly I’m not a video games person and had no idea that “case” was the same thing as...rating? and also I have no idea what an E rating is, I don’t recognize that one from movies.)
“A video game sitting against a wall, rated E, grey rimmed, metal temples along a concrete river with silver gates, chests, and floating gold keys. HD photo, macro.” I don’t think it super knows what you want here…
A “case” in this context is the plastic clamshell that holds/protects the disc when not in use (DALL-E thinks this instead means some sort of container found within a video game environment).
The E rating (for “Everyone”) is similar to the G rating for movies.
What it should be creating is this below (a video game case) … XD lol:
https://ibb.co/9TtJbqJ
I hypothesise that the more details a prompt contains, the more likely DALL-E is to throw a wobbly and produce something almost totally random. But honestly, I’m very impressed with the outcome of most of these prompts. The picachu eating a hamburger is the only one of the above that really “failed”, and a couple of the outputs picked up about half the requested details.