NoesisGUI and AI prompting
Recently picked up noesisgui to use for UI creation in my game, and really enjoy the WPF styled design and data control vs unity's built in UI systems. I've been having decent success with generating controls using AI, as it has a much better understanding of xaml than I do (i absolutely suck at figuring out how to properly align/size/stretch my ui elements in xaml). There are some problems im running into however, a lot of times it gets confused between WPF and noesis, and tries to do things that don't quite work the same way. one super simple example is using doubles instead of floats, where as a more complex problem might be referencing a wrong class, or using an existing class the wrong way. I'm going to be experimenting with passing documentation from https://www.noesisengine.com/docs/Gui.C ... archy.html into my prompts selectively/as needed when generating new user controls, but im wondering if there's anyone else out there who's been utilizing, or thinking about using aimodels to generate UI as well, and would like to hear their experience/ideas? Or perhaps if the developers have some suggested guidelines or ideas on directing my prompts to be more in tune with how noesis handles certain aspects of code compared to WPF, that could be insightful into getting better output that i don't need to fiddle with as much.
I've got a friend who's been extremely productive in working on UI for his game, and ive been quite impressed by the progress he's making (HTML5/react game), So I know the models are capable enough to do the work, but i think their understanding of noesis isn't complete / defined well enough in their training data to be competent in using it the same way (hence passing appropriate documentation along with the desired output)
Anyways, looking forward to and appreciate any ideas/responses./discussions on the topic
I've got a friend who's been extremely productive in working on UI for his game, and ive been quite impressed by the progress he's making (HTML5/react game), So I know the models are capable enough to do the work, but i think their understanding of noesis isn't complete / defined well enough in their training data to be competent in using it the same way (hence passing appropriate documentation along with the desired output)
Anyways, looking forward to and appreciate any ideas/responses./discussions on the topic
-
- KeldorKatarn
- Posts: 243
- Joined:
Re: NoesisGUI and AI prompting
From a personal experience as a almost 15 years WPF developer: Try to figure this out yourself and only use AI to give you pointers on what to look at and how to solve problems.
XAML has intricacies that will massively bite you in the butt unless you really know what you're doing. AI will produce something that way work, but then ends up having little bugs or inefficiencies that will cost you more time fixing than learning how to do this yourself in the first place. Using AI for XAML and control design when you don't know what the AI is actually doing is a recipe for disaster. Just my 2 cents.
XAML has intricacies that will massively bite you in the butt unless you really know what you're doing. AI will produce something that way work, but then ends up having little bugs or inefficiencies that will cost you more time fixing than learning how to do this yourself in the first place. Using AI for XAML and control design when you don't know what the AI is actually doing is a recipe for disaster. Just my 2 cents.
Re: NoesisGUI and AI prompting
Well, I wrote out a long post explaining my process but the forum decided to auto log me out after an hour and a half of writing 🙁.
In short, not using the 'vibe code' methodology, I've got enough experience under my belt (I believe) that I can recognize things that wont scale/work correctly. I analyze the output code I'm getting and make sure it fits my needs, and if not I'll either reprompt or make the implementation myself. Is there a scenario you foresee where I'll hit a wall? Losing my last post has me a bit burnt on writing more at the moment. The main value I see is in generating quick mockup UI's I can modify and start coding for without the need to design every panel/child control. The UI's im making aren't super complex by any means, just simple things like overworld map, containers/inventory, tabview styled menu, information panels, etc.
Heres a sample of a generated control:
These are the kind of 'mockups' im looking to generate. Note here I havent passed any documentation in my prompt, so its ended up hallucinating a 'spacing' property for the stackpanel class. I believe this is because of a 'contextual' ambiguity in the model, as such, if we refer to the windows docs, their stackpanel class does utilize a spacing property:
https://learn.microsoft.com/en-us/uwp/a ... el-spacing
In short, not using the 'vibe code' methodology, I've got enough experience under my belt (I believe) that I can recognize things that wont scale/work correctly. I analyze the output code I'm getting and make sure it fits my needs, and if not I'll either reprompt or make the implementation myself. Is there a scenario you foresee where I'll hit a wall? Losing my last post has me a bit burnt on writing more at the moment. The main value I see is in generating quick mockup UI's I can modify and start coding for without the need to design every panel/child control. The UI's im making aren't super complex by any means, just simple things like overworld map, containers/inventory, tabview styled menu, information panels, etc.
Heres a sample of a generated control:
Code: Select all
<UserControl xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
Width="500" Height="300">
<Border Background="#222" Padding="10" CornerRadius="8">
<StackPanel Spacing="10">
<!-- Character Portrait and Dialogue -->
<StackPanel Orientation="Horizontal" Spacing="10">
<!-- Placeholder Portrait -->
<Border Width="80" Height="80" Background="#444" CornerRadius="4">
<TextBlock Text="NPC" Foreground="White"
VerticalAlignment="Center" HorizontalAlignment="Center"/>
</Border>
<!-- Dialog Text -->
<TextBlock Text="The dialogue box displays the words the character is currently speaking."
TextWrapping="Wrap" Foreground="White" Width="380" />
</StackPanel>
<!-- Scrollable Response Options -->
<ScrollViewer Height="150">
<StackPanel>
<Button Content="Do you need any help?" Margin="0,5,0,0" Padding="8"/>
<Button Content="Tell me more about yourself." Margin="0,5,0,0" Padding="8"/>
<Button Content="Why are you here?" Margin="0,5,0,0" Padding="8"/>
<Button Content="Goodbye." Margin="0,5,0,0" Padding="8"/>
</StackPanel>
</ScrollViewer>
</StackPanel>
</Border>
</UserControl>
https://learn.microsoft.com/en-us/uwp/a ... el-spacing
Re: NoesisGUI and AI prompting
I'm really sorry to hear that. I'm reaching out to IT to see if there's anything we can do to improve the session timeout settings.Well, I wrote out a long post explaining my process but the forum decided to auto log me out after an hour and a half of writing 🙁.
Yes, that's because the documentation you're referencing is for UWP. Since we're using WPF as our base, the StackPanel class differs slightly. The WPF version doesn’t include a Spacing property, here's the correct reference for that:These are the kind of 'mockups' im looking to generate. Note here I havent passed any documentation in my prompt, so its ended up hallucinating a 'spacing' property for the stackpanel class. I believe this is because of a 'contextual' ambiguity in the model, as such, if we refer to the windows docs, their stackpanel class does utilize a spacing property:
https://learn.microsoft.com/en-us/uwp/a ... el-spacing
https://learn.microsoft.com/en-us/dotne ... stackpanel
Who is online
Users browsing this forum: Ahrefs [Bot] and 40 guests