I use GPT to create animated SVGs from static ones - in t3 I currently have to paste the xml into the chat and then save the output to view it. Would love to see uploads supporting svg and an svg preview option for the chat. But make sure I can still see the markup.
Turquoise Chocolate
+1 on this, was about to open a request about this very thing and saw it already existed!
Seeing that Claude and ChatGPT have builtin SVG viewers, I feel like it makes sense we get one with the T3.chat level of polish!
Coral Cucumber
+1 on this. Claude has an inbuilt svg viewer for displaying svg code it generated. This would be super useful
Magenta Andromeda
I also would find this feature really useful as I use SVGs all the time in my work. Every single AI UI is TERRIBLE at viewing SVGs or flat out doesn’t support it. Plus, it would be even better if users could select an element from a generated SVG and prompt the AI based on that section in context too. (i.e. move the element I just selected in the SVG up)
Mark Florkowski
Team
What you described with selection should work in theory right now. Not sure how the models will respond, but the selection quoting does exist.
regarding rendering inline, it’s a bit difficult because we need to validate they are “safe” , and svg render behaviour varies pretty significantly between browsers. Will look into this again though net time I get a chance
Magenta Andromeda
Yes, I know from personal experience that SVGs can a little weird but they are very useful. It would be much more efficient for my workflow if I could view SVGs generated by the AIs directly in t3.chat. Also, when I asked about the section quoting I don’t just mean by selecting the text of the code element I want. Instead I would really like to see the SVG loaded and be able to click the elements, rather than the raw text, and make the AI reference that part needs to change in my next message to it.
Log in to comment and vote
Comments7
Moccasin Meerkat
I use GPT to create animated SVGs from static ones - in t3 I currently have to paste the xml into the chat and then save the output to view it. Would love to see uploads supporting svg and an svg preview option for the chat. But make sure I can still see the markup.
Turquoise Chocolate
+1 on this, was about to open a request about this very thing and saw it already existed!
Seeing that Claude and ChatGPT have builtin SVG viewers, I feel like it makes sense we get one with the T3.chat level of polish!
Coral Cucumber
+1 on this.
Claude has an inbuilt svg viewer for displaying svg code it generated.
This would be super useful
Magenta Andromeda
I also would find this feature really useful as I use SVGs all the time in my work. Every single AI UI is TERRIBLE at viewing SVGs or flat out doesn’t support it. Plus, it would be even better if users could select an element from a generated SVG and prompt the AI based on that section in context too. (i.e. move the element I just selected in the SVG up)
Mark Florkowski
What you described with selection should work in theory right now. Not sure how the models will respond, but the selection quoting does exist.
regarding rendering inline, it’s a bit difficult because we need to validate they are “safe” , and svg render behaviour varies pretty significantly between browsers. Will look into this again though net time I get a chance
Magenta Andromeda
Yes, I know from personal experience that SVGs can a little weird but they are very useful. It would be much more efficient for my workflow if I could view SVGs generated by the AIs directly in t3.chat. Also, when I asked about the section quoting I don’t just mean by selecting the text of the code element I want. Instead I would really like to see the SVG loaded and be able to click the elements, rather than the raw text, and make the AI reference that part needs to change in my next message to it.
Mark Florkowski
Ah I see.