Fix readme

This commit is contained in:
Jeroen Domburg 2023-04-23 14:29:58 +08:00
parent 2165990cfd
commit 40d6fa237e
2 changed files with 9 additions and 49 deletions

View file

@ -1,37 +1,16 @@
* Workflow
Uploading:
- html/js: select image, crop, resize, upload to server
- upload php: use C program to convert to EPD.
Spit out preview (?), add epd to database w/ timestamp
### What is this?
epd:
- Wake up
If Internet:
* fetch index JSON file (contains OTA plus timestamps of last 10 images)
* Update files to make internal store match
* Update list of image shows: set to 0 for new images
- Take list of image shows; get lowest show count; show newest image with that count.
This is the firmware, PCB artwork, schematics and case design for a WiFi-enabled color picture frame.
Flash:
Other shit: 0x10000
OTA1: 0x150000
OTA2: 0x150000
left: 0x150000
### Contents
img size 134464
11 images -> 10 150000
- case - OpenSCAD program to build the case. Change the 'render' variable to render the various parts.
- www - PHP and C program running on the server that is used to upload/fetch the pictures.
- firmware - ESP32C3 firmware. Make sure to run menuconfig to point this at your server.
- pcb - Schematics and PCB artwork, in Kicad format.
### More information
Case uses M3 inserts: height 4mm dia 5mm
Corresponding screws: M3*6mm
Colors:
- Grab picture in 'normal' situation (daylight, curtains, ...)
- Open in Gimp
- Image -> precision: 16-bit, linear
- Filters -> blur (to get rid of grain)
- Curve -> make darker colors a bit more dark (so it mixes more white in there, making end image lighter)
- Filters -> levels, select black level, select white level
[Article on Spritesmods](https://spritesmods.com/?art=wifipicframe)

View file

@ -1,19 +0,0 @@
For PCB:
XR1151 - 800'ish mA, 65uA quiescent -> 3.5 year
ESP32C3 - need to test if works with same code, otherwise NP
Article:
Color Floyd-Steinberg is a PITA. Difference colors? LAB colors, euclidian thingy. That sucks,
pink white and blue-ish white is the same difference as red and blue. Initial modification:
multiply A and B differences (=hue difference) with sin(L). This makes those differences less
important. Great results. But WtF is up with the shitty 'official' way? Seems there's multiple
ways: latest and greatest is E2000. Actually not that much different than my hacky solution.
Non-linear color space: one pixel of (0,0,0) and one of (64,64,64) don't look the same
(are not as bright) as two of (32,32,32). Luckily (linearized) RGB works.
Stucki dithering?
https://mathematica.stackexchange.com/questions/7483/how-to-calculate-mix-of-4-colors-defined-in-cielab-lab-model