hey there! I was unable to add my readmes' using git checkout (it is very late at night), so I have added them all here. sorry about this! :) they go in order from most recent - oldest and there are 6 entries for the 6 purview holder things.)
##Seccond round of testing - FINAL
this is my final: here are my official parameters:
Hair color --> opacity of the ears face width--> length of the whiskers gender--> fluffy for boys, sharp and angular for the girls eye color --> eye color skin color --> deepness of skin color
some things I had to change from the last round of training:
- reduced the number of eye colors to choose from, as the AI can barely see the difference between two.
- made the fur correspond to the skin color of the person I was testing, because hair was very iffy in terms of the AI learning how to do that to.
- opacity of the ears directly corresponds to the darkness of the hair now.
Some aestheitc changes I made:
I made added shading detail in the female cats, making them look a little 3D and angular. I could not find a way to do this super nicely for the men.
Choice in images:
All three have a different narrative, and are not related in any way
Crazy cat people --> this one is relativly self explanatory, crazy cat people wearing crazy cat mask faces! Kardashians --> pussy power, powerfull women, enough said. Past US presidents --> it'd be nice to view the past presidents as nice fluffy aproachable people rather than big scary men in suits. It represents that everyone has a soft side.
My overall thoughts on the final product:
I do like it, I just wish I had allowed myself more time to work out the kinks in the design and the testing. As I am a perfectionist this is not enough for me, but I think it is adequate/pretty good for the time contraints I have had thoughout project 2 (from other subject, asignments, and work). I like my ps1 better, and I think the two link together well.
##First Round of Training:
Things I have trained:
hair color --> changes the color of the cats fur (i.e. people with dark hair will have black or brown - people with red hair have pink)
Face width --> the length of the cats whiskers (i.e. much like in nature the wider the cat the wider the whiskers).
Gender --> People who present as mascualine become cute and fluffy cats
Facial Hair --> men with facial hair have darker ear inners/eat opacity
Eye color --> I am training at the moment to blue, green, brown, and dark eye colors. This is in a very 'beta' stage. i.e is not working so well. In my next iteration I will definitly be changing the colors to more like: 'light, medium, and dark' to try and confuse the AI a little bit less
The other thing which is currently in beta is the hair color. As kim kardashian certainly isnt ginger - but the Ai believes she is. I will likely be changing and training further.
If I have time/ mad passion I will defintily add some more sliders which can be further customisable to the Individual which is mapped.
Choice in images:
All three have a different narrative, and are not related in any way
Crazy cat people --> this one is relativly self explanatory, crazy cat people wearing crazy cat mask faces! Kardashians --> pussy power, powerfull women, enough said. Past US presidents --> it'd be nice to view the past presidents as nice fluffy aproachable people rather than big scary men in suits. It represents that everyone has a soft side.
##fourth entry:
I am experiementing with testing today, I was having some issues with getting the testing to work on the images I had selected. They would train fine, and the quiz would be fine, but they would reset to the default face and be all the same in the faces section. It turns out there was an issue with merging part three. Tom fixed it for me.
The mappings I have now are: tilt, character type (the men are fluffy), whisker length, color, ear opacity, eye color. I have changed the color from black and white to natural hair colors, in order to have some diversity when mapping hair color. I am not sure how I feel about it though.
##Third Entry:
In order to make my mak more mask like, I deleted every second vertice from the chin mapping. This created a more structured look which I feel made it look more mask like. This also got rid of the annoying chin point. In addition to this, I added names to my sliders. I have tilt, whiskers and mouth, gender, and glasses. I ended up getting rid of the mouth, and replacing it with a for loop of just the top lip. This makes them look a little creepy cat/human like, which I am pleased with. I have not yet linked the sliders but I will asap.
The other thing I did was experiment with pictures, I added the kardashians in which I like, but I will probably revert back to the origional image for tersting purposes. Soley becaue the kardashian picture is far away, and harder for me to see.
##Second Entry:
I decided to go with a a geometric theme, as I felt this would be easier to implement using the points given from the mapping. The only thing I do not like about this is that the chin is super long for a cat. I would prefer for it to look more like a mask rather than a weird cat face imposed on a human. I can also not nolonger like the mouth - as thr line does not line up with the nose (which I made using a for loop).
I also re made the eyes. I did this by using a for loop to circle around the points, and then I adjusted scale to make them big and exadgerated like cats eyes. The ears are based on the eyebrows. I like having almost everything mapped to the verticies as it looks more dynamic in relation to the face.
At this point, I only one slider, which is tilt at this point. And also whisker length/mouth length.
##First Entry:
Importing my values from ps1 into code for ps2. Have added in some aspects of my previous design, have not implemented the changes yet. I did have to hard code as the scale was different from ps1. I redid all my code.