Welcome, Guest. Please login or register.

Author Topic: Found a scan doubler ( who would like one ) ?  (Read 17623 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline nBit7

  • Full Member
  • ***
  • Join Date: Jul 2007
  • Posts: 143
    • Show all replies
Re: Found a scan doubler ( who would like one ) ?
« on: October 30, 2007, 12:11:13 AM »
Maybe someone can find or make up a graphic that shows a 256 8bit single colour gradient next to the equivalent using only 4 and 5 bits.

That way people can see the difference for them self without having to resort to fiddling with screen modes.
 

Offline nBit7

  • Full Member
  • ***
  • Join Date: Jul 2007
  • Posts: 143
    • Show all replies
Re: Found a scan doubler ( who would like one ) ?
« Reply #1 on: October 30, 2007, 12:50:53 AM »
To be able to reproduce the full 16M colours would require using more than 24bits (3*8bits) at the analogue to digital converter (ADC).
This is because it would near impossible to line every analogue step with the corresponding digital.  For example suppose the analogue Red signal with 256steps (8bit) went from 0Volts for off to 1 volts.  Then each step will be 1/256 Volts or 3.9mV (0.0039V).

colour value = Analogue voltage level
-------------------------------------
0 = 0V
1 = 3.9mV
2 = 7.8mV
...
253 = 992mV
254 = 996mV
255 = 1000mV  (1V)

If the DAC was only looking at a range of say 0 to 995mVolts
it is clear that it would not correctly detect all the levels.  This is even before we consider all the various types of distortion.

If we sampled with say 16bits per colour and make the digital capture range slightly larger than the analogue then we can capture the full range of colours.
 

Offline nBit7

  • Full Member
  • ***
  • Join Date: Jul 2007
  • Posts: 143
    • Show all replies
Re: Found a scan doubler ( who would like one ) ?
« Reply #2 on: October 30, 2007, 05:27:59 AM »
Quote
Cutting out the 0-255 color diagonal gradients and viewing them with PPaint, there is a loss of color (255 color gradients aren't perfectly smooth), however they look better than I expected, and the color loss in this case could be down to PPaint image handling/conversion process. The 256 color red sphere looks amazing since the banding is less perceptible.


remember that the gradients are restricted to 256 different colors at once.  But those 256 colours can be any out of the 24bit.

For 24bit colour RED GREEN and BLUE each can be set at 256 (0-255) levels (intensities)

If you were to use a gradient that used only one of those colours (eg. Black to RED, but not Black to RED to White).  In this case you would grantee that you didn't go over the 256 different colours at once limmit.   Then it can be displayed in 256 colour AGA with no loss what so ever, as there are only 256 red levels (in 24bit graphics) and AGA can display every one of them.  
An ECS or OCS amiga would only be able to display 16 levels of that gradient.  16levels leads to noticeable banding.

In an example where you had a gradient that went from Black to RED to White.  You have a situation where you can have 256 levels getting to full red then another 255 levels getting to white.
RED to White is made by keeping RED at 255 and bringing up GREEN and BLUE together.