File talk:ZX Spectrum standard palette with 8x8 dithering.png

Higher DIM values than 205?
Hi! , 4throck uploaded a personal version of this old chart¹. In this process, all the 205 dim values (out of 255 on our 24-bit systems) were replaced with quite brighter 238 values (besides the lack of further Png compression and Png metadata preservation, typical of a quick edition).

This looks overkill to my trained eye:
 * 1) I used to 'code' and 'draw' during several years on these television-connected microcomputers back in the early 1980s, so indeed, I can tell that the original dim values were not artist-friendly at all, and quite more complicated that setting e.g. a 205 value in the lower part of the palette;
 * 2) I remember many inconsistencies on the original ZX Spectrum hardwares (16/48K and above) and with their cheap displays (home televisions with different effects on the brighness, saturation and contrast), I agree; but 238 looks quite too bright in my recollections of using the original line of Speccies myself;
 * 3) YouTube shows many original hardwares in action: I never spotted this dramatic gap after hundreds of these specific viewings;
 * 4) therefore, I wonder if 4throck's PAL installation was as correct as intended: as it is said, I suspect e.g. a heavy-handed brightness setting (deceptive manual approximation), or/and faulty hardwares (biaised related 'conversion').

Our comments: We have a pixel art tendency to darken even more these 7 dim colors (as if they were mid-grays): this eases the neo-artists (since the medium gray 192 would be ideal to them without this limitation), at the expense of some technical faithfulness. Please bear in mind that the varied machines and display settings cannot provide less than a series of preferences to the retro scene (the most pernickety retro pixel art galleries provide a set of popular palettes). Well, this discussion may be endless, depending on what we consider the main priority in today's renditions…
 * 4throck's: « converted to approximate PAL gamma - better representation of non-bright colours as displayed on real hardware »;
 * dpla's, as a disclaimer: « the accuracy of the dim colors is questionable; 205 in the 3 RGB components seems to be widespread, while the darker 192 values are seen on Wikimedia ».

However, I reverted the palette to its original upload without feeling remorse, 4throck:

¹ I never came across this ambiguous suggestion (i.e. any similarly undocumented choice based on a very faint dimming). I assume that even Sinclair Research would have considered this ineffective 'bright 0', a material flaw. 205 is far from being universal, admittedly, but it's still brighter than the average gray (191 or 190 according to gamma-related specialists – 192 [= bitwise-corrected or 1-256-scaled '255-32'] is preferred for RGB convenience). 238 (~ 239, the well-known [accurate '255-16'] step) cannot represent a viable solution from any standpoint, in 2021 or 1982. Let us know why in practice, otherwise. Bye!

-- d⠚p⠖ l ⠃a⠒ .fr 01:54, 1 September 2021 (UTC)