Juliu retweeted
Got first 2 lovely reviews!
(For some reason not yet showing on Google Play, but visible for me.)
#reviews #googleplay #5stars #uxdesigns #tinyUX #wireframing
Juliu retweeted
Yay, made the #hackernews homepage!
Juliu retweeted
Website live for TinyUX: tinyux.app
Just tap tap tap to quickly create low-fi wireframes anywhere, anytime!
#UX #Wireframing #taptaptap #Tiny
Juliu retweeted
Next app update will include the save-copy-clipboard-as-component feature.
Have it working, but need to add the remove-component feature too.
#ux #tiny #Components #stars
Juliu retweeted
Reviving this Twitter account. Making good progress on UX app.
Juliu retweeted
"Forest. Painted by Julius Huijnk"
"Forest. Illustrated by Julius Huijnk"
I like 'my' illustrated style more.
Btw this is with only 30 steps, 50 is probably better. I should be able to improve on it based on the seed.
#stableDiffusion
Juliu retweeted
Google taking its time for the review..
Meanwhile, I'm working on merging face elements. There is UX challenge on how to 'break' if desired.
First step is auto-merge if same type is connected, like mouth here.
Juliu retweeted
New feature is that you can mix to discover new shapes.
#silly
Juliu retweeted
Another experiment with live streaming illustrating on Twitch, I'm starting right now at this link:
twitch.tv/juliushuijnk
Juliu retweeted
Replying to @PokeyRule @GoCursorless
Couple of resblocks with linear layers. So stacked convolutional layers Quite simple architecture.
I have no experience with RNN decoding . Would be cool to learn that too, but I have the feeling I'm getting quite close with this approach. So pushing this ahead first.
Juliu retweeted
Replying to @Juliu @GoCursorless
In essence this works. I'm mostly working on improving it, handling larger grids. etc. Also tried different ways to encode the size of components in the output-nodes of the net.
Juliu retweeted
Replying to @PokeyRule @GoCursorless
Did write an article once on the need for a UX source language. medium.com/proof-of-concept/…
But for the model I'm now using what you might describe as one-hot-encoding per cell in a grid. Each cell has a group of output-nodes, each node represents a component (img, button, etc).
Juliu retweeted
Tried a bunch of ways to wrap my hand around it. Ended up with top-lef-corner as anchor idea. A wireframe is unique in that we can assume components don't overlap.
For grouping in a menu, cards, etc, some more stuff is required, but I got ideas for that too.
Juliu retweeted
Hier het artikel:
linkedin.com/pulse/sir-helpa…
Anti-cyclish gedacht (en bij gebrek aan geduld) geplaatst in het weekend, zodat de feed verder vrij rustig is daar.




















