Closed chstock closed 7 months ago
Thanks for working on this, @chstock! I really like the look of the image, and I like that you used code to generate it (as opposed to existing images that may come with copyright restrictions). Would be great to discuss in our meeting next Wednesday. Just a few comments to start:
/man/figures/
. Example: https://github.com/tidyverse/dplyr/blob/main/man/figures/logo.png. It would be nice to follow this precedent for logo.svg
and the R or Quarto script that generates it.logo.svg
, PNG files of various sizes, and only the most necessary information about the images was generated.Many thanks for your review, @wlandau! Yes, let's discuss in more detail in the team meeting next Wednesday.
A few quick responses:
- Typically, packages have logo files in
/man/figures/
. Example: https://github.com/tidyverse/dplyr/blob/main/man/figures/logo.png. It would be nice to follow this precedent forlogo.svg
and the R or Quarto script that generates it.
The image file (logo.svg
) indeed is in man/figures/
, see here. I am happy to put the script there (or elsewhere) too. I suspected it may complicate documentation or package building if I put it there, but I haven't tried.
- Could we remove the intermediate images, rendered Quarto HTML, and Adobe Illustrator files? In the latter case, the file format might be copyright-restricted. I usually prefer to just have the main
logo.svg
, PNG files of various sizes, and only the most necessary information about the images was generated.
Yes, happy to remove the intermediate images and Ai files. To be discussed maybe that the R script doesn't generate the final image, there are some final touches wit Ai that I believe would be difficult with R alone. This information and the possibility to fully reproduce the image and make quick modifications would be lost.
- I like that you used a color scheme tool to pick the colors, but I wonder if orange or dark yellow would look friendlier than red.
Sure, let's discuss... We can also pick another color scheme.
For what it's worth, the app at https://connect.thinkr.fr/hexmake/ helps turn an image into a hex logo. Maybe this could replace Ai?
I updated the hex sticker:
inst/hex_sticker/
(to be discussed),logo.svg
is in man/figures/
, and png
s) have been added to inst/hex_sticker/favicon/
.Happy to further modify the colors and/or font, or implement other requests.
The current version looks like this:
Thank you for the changes, @chstock. I really like the improvements. A few follow-up requests:
brms_mmrm_hex.html
file and the generated images from the inst/hex_sticker/
? Installed package size is always at a premium, and I think we only need the source Rmd file there.pkgdown
sites, favicons are typically generated by pkgdown::build_favicons()
and stored in pkgdown/favicon
. Could we transition the favicons there?Thanks, @wlandau. Yes, this all makes sense to me too...
- Could we remove the
brms_mmrm_hex.html
file and the generated images from theinst/hex_sticker/
? Installed package size is always at a premium, and I think we only need the source Rmd file there.
Done in https://github.com/openpharma/brms.mmrm/pull/89/commits/582825a5c598d98c6d6b45c1f787541f617a7119.
- For packages like ours which have
pkgdown
sites, favicons are typically generated bypkgdown::build_favicons()
and stored inpkgdown/favicon
. Could we transition the favicons there?
Done in https://github.com/openpharma/brms.mmrm/pull/89/commits/f278e88ccd026ee6f4d06c427b295ad05dea3e06.
- For simplicity, it would be nice if the border color agreed with one of the colors from the main graphic. The green border looks faint when I run the current image through color blindness simulators (https://pilestone.com/pages/color-blindness-simulator-1).
Done in https://github.com/openpharma/brms.mmrm/pull/89/commits/1f66d9d05b6938997f88d8a30684109a6b9f955c.
This is the updates image:
Again, happy to make modfications (w.r.t. color, font, etc.)
I am proposing a hex sticker:
The sticker contains a subplot that was created based on a simulated data using
{ggplot2}
and{ggdist}
, particularlyggdist::stat_halfeye()
. I thought this plot may well convey the key features of the Bayesian MMRM (experimental, comparative analysis, continuous endpoints, discrete time variable and probabilistic estimates). The color scheme was taken from Huemint.The initial figure was further processed by
hexSticker::sticker()
, from which I didn't use one of the standard svg or png output files but saved a pdf. The pdf was then minimally modified further in Adobe Illustrator. The final files (hex_sticker1.svg/.png/.pdf
) were exported from Adobe Illustrator. This workflow was convenient to get quite clean image files.All files (allowing reproduction or modification) are in a new folder
inst/hex_sticker
which I added to.Rbuildignore
.The sticker was added to the
README
file (usingusethis::use_logo()
) andREADME
was rerendered.Just let me know what you think and whether you would have certain preferences for a hex sticker (subplot, colors, font, ...). I can then try to accomodate this accordingly.