We want to make sure you get the best - we want to make sure of this even if it comes at a cost to ourselves. After investigating StableDiffusion we started to have a look at DALL.E, run by the creators of Chat GPT, to see if that would make a difference to the image outcomes.
Silverley Event Visualisation vs AI Renders
Below is the image we were trying to create in DALL.E:
We put the following prompt in:
Old Billingsgate Market from the outside looking towards the arched entrance of the building. A Formula E car should be outside with a step and repeat board showing the Formula E logo. There should be a red carpet outside next to the car and the red carpet should lead up to the entrance. There should be candles on the steps to the right and left of the red carpet. It should look like a 3D generate
Unlike StableDiffusion, DALL.E costs to use. It has a limit on prompt length and cut our prompt off early - but it did produce the images below.
The images above are noticeably better than those by StableDiffusion: they have a red carpet, a car and in the image on the far left there are bollards that look like candles - but, even were bits to be edited, there is still a huge amount of work to be done to match the image to the prompts - in fact, it would be easier to take a snapshot outside of the venue and photoshop it. In terms of the 3D visuals renders therefore, there's a lot more work to be done until they're pitch ready.
Silverley Event Visualisation vs AI Renders Round 2
We had another go - the arched buildings appear in all of them, as does the colour red.
Silverley Event Visualisation vs AI Renders Round 3
We tried one last time, using the phrase "digital art" as recommended for producing high quality images:
Again, we manage to get some cars, something that looks like a Formula E track, the colour red and the high arches of Old Billingsgate.
What does this mean?
We can't write off the AI image generator completely - the image below is also produced by AI from the prompt:
a room in the versailles palace decorated for a luxurious dinner
and while things will start to go awry if we then become too specific with the chair type and colour, what the candler holds should look like or the type of lighting, the images themselves provide good fodder for a mood board.
Conclusion
Silverley Visuals is a better choice than an AI text-to-image generators. As humans, we can not only read and create your brief, but we can ask further questions, collaborate with you and ask for further visual clarification - helping you refine your ideas while we breathe life into them. Secondly, while DALL.E and StableDiffusion are much more advanced than any previous text to image generator created by AI, the technology is still developing. In contrast, our designers have years of architectural experience using industry leading design software.
What we can do for you
Head to our What We Do section to find out more about how we can help you, or book an online consultation to talk through your wedding and event visualisation needs.
Comments