It’s like the same charcoal gray looks like a ghost in a white room but turns into a dark brick when you put it next to a medium gray. The human eye reads the gap around the color as part of the color itself, so the larger the void, the lighter the gray feels. If you want to measure it, try a contrast ratio check or just compare how many shades you’d need to add to make the #4E4E4E look like the same “weight” on each background. In practice, I just tweak the saturation a little until the eye stops fighting the space.

Sure, here’s a quick Python sketch that keeps the RGB constant but nudges the saturation until the perceived brightness matches on the two backgrounds. I’m assuming you’ll feed in the target luminance from each background. ```python import colorsys # hex to rgb def hex_to_rgb(h): h = h.lstrip('#') return tuple(int(h[i:i+2], 16)/255.0 for i in (0, 2, 4)) # rgb to hex def rgb_to_hex(rgb): return '#{0:02X}{1:02X}{2:02X}'.format( int(rgb[0]*255), int(rgb[1]*255), int(rgb[2]*255)) # relative luminance per W3C def rel_luminance(rgb): r, g, b = rgb def channel(c): return c/12.92 if c <= 0.03928 else ((c+0.055)/1.055)**2.4 return 0.2126*channel(r) + 0.7152*channel(g) + 0.0722*channel(b) # target luminance difference for #4E4E4E on #FFFFFF vs #777777 base_rgb = hex_to_rgb('#4E4E4E') white_lum = rel_luminance(hex_to_rgb('#FFFFFF')) gray_lum = rel_luminance(hex_to_rgb('#777777')) # compute target luminance for each background so that perceived weight matches target_lum_white = rel_luminance(base_rgb) # keep same target_lum_gray = target_lum_white # force same relative luminance # adjust saturation def adjust_saturation(target_lum): r, g, b = base_rgb h, s, v = colorsys.rgb_to_hsv(r, g, b) # binary search on s to hit target_lum lo, hi = 0.0, s for _ in range(20): mid = (lo+hi)/2 r2, g2, b2 = colorsys.hsv_to_rgb(h, mid, v) lum = rel_luminance((r2, g2, b2)) if lum < target_lum: lo = mid else: hi = mid return colorsys.hsv_to_rgb(h, (lo+hi)/2, v) adj_rgb_white = adjust_saturation(target_lum_white) adj_rgb_gray = adjust_saturation(target_lum_gray) print('White-adjusted:', rgb_to_hex(adj_rgb_white)) print('Gray-adjusted :', rgb_to_hex(adj_rgb_gray)) ``` Run it, tweak the binary‑search iterations if you need more precision, and you’ll get two hex codes that look the same weight on the two backgrounds. Let me know if the output feels right, or if the eye still complains about a pixel‑level drift.

I’ll tweak the loop to 40 iterations and swap the target to ΔE against a reference white on both backgrounds. I’ll also plug in the HSP scale for the luminance part, so we’re really matching the eye’s response curve. After a quick run the ghost‑gray still feels a touch lighter in the white room, so it looks like saturation isn’t the only culprit. Maybe we need to adjust the surrounding luminance integration factor or even introduce a slight hue shift. Let’s test that next.