This becomes apparent with more colors involved - generally "averaging colors" is not a trivial thing. Then you would be contrasting against a gray background, yet that is never the case. Imagine that you collect 10 angles where the background is pitch black, and 10 where your background is clean white. However, there seems to be a slight issue with your methodology. I haven't uploaded the data, but you can collect your own with relative ease, just use like OBS or something if you want to.įirst, I really love your approach and the work you have put into this. big angles) more would impact the decisions, but I'm overthinking this enough as it is.ĮDIT: Since a solid chunk of you are interested in the actual specifics, here's a link to a Github repository with the script and some basic instructions to get started. It'd be cool to see how considering multiple dominant colors per frame, or weighting specific important frames (e.g. I'm not a color person, so it's entirely possible this is a super bad choice, but idk. From here, I just inverted the color and maximized the saturation and value of the color to make it as bright as possible. If I was analyzing multiple maps, I computed the colors for every map, then averaged all of those averages together (this makes sure that each map equally contributes to the average "game color"). This color was then collected for every frame in the map and averaged to get the overall "color of the map". From there I used OpenCV to load each video frame by frame, then used PIL and K-Means to extract the dominant color from a specific cropped section of the image (usually a square crop around the center of the image). I collected the videos by running around in customs without a crosshair and recording the video with OBS you could get colors from game files but I'm not nearly that smart. Long story short, it's a bunch of OpenCV, PIL, and NumPy thrown together in an abomination to all software engineering. Ok, a bit of a crash course in how I actually wrote this script. The data I gathered was pretty sparse and it'd be interesting to see how these choices fare when considering smokes/util/etc., but I digress.Įnd of the day I may still suck at this game, but it's not because my crosshair is bad. In any case, the crosshair colors I found work well enough, so it'd be cool to see if anyone else can find a better solution based on this. However, the blue trend is enough to noticeably shift the overall color to a sky blue, which is cool to see. Icebox and Pearl are the only real exceptions to this rule, using bright yellows/reds respectively. There's a couple of interesting patterns here Cyan/Blue is a very common choice of color, especially for Ascent, Bind, Breeze, Fracture, and Haven. Split: may or may not have gotten too lazy to collect data here Too many custom games and a quick Python script later, I came up with a list of crosshair colors that contrasted as much as possible with each map, as well as an overall crosshair color that attempts to work for every map. So instead of playing the game and finding my preferences like a normal person, I recorded myself in customs looking at every angle I could think of on every map (except Split RIP) and wrote a Python script to look through that video find the best crosshair color for me. When Patch 5.04 dropped and crosshair craziness happened, I decided to try and find the best possible crosshair color in Valorant based on video from actual Valorant gameplay.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |