Forums

I've noticed on some games recently that the colours are not quite right. In particular, Fantastic Dizzy uses a lot of browns which look much worse emulated than on-screen, and when playing through Micro Machines yesterday I kept thinking "that isn't right".

So the obvious solution is... a homebrew program which will display large(ish) blocks of any colour on-screen (16 at a time would be useful and about the right size for each) which I will grab and attempt to smooth out to get "good" colour values. The problem is, I imagine the end result will still be quite variable. If I had a composite or RGB connection I expect the colours would be slightly different; then there's the issue of monitor gamma. My TV card does a pretty good job of converting between TV and monitor gammas though.

Can anyone oblige with the code? I really need a book (or doc) to work from before I can even consider coding anything much myself.

Maxim

Quote

> So the obvious solution is... a homebrew program which will display large(ish) blocks of any colour on-screen (16 at a time would be useful and about the right size for each)

Here's a program I made for you to do that.
http://www.finalburn.com/colcheck.zip

Many thanks to Dave, Charles and Bock for their various implementations. Here's that image:

http://www.webtribe.net/m/maxim/temp/coltest1.png

Taken from Charles' first program, mainly because his has the largest blocks of colour. It was all nicely aligned until I realised I had to rearrange it to get the RGB separation as shown, which is why it's a little misaligned. I hope the meaning of the numbers is clear. However, here are some things to consider:

- My TV card is quite variable. Artificial solid blocks of colour do not come out as well as natural images. You'll probably have noticed the same if you ever tried to video your SMS, although the effect is much worse on a VCR.

- Charles also wrote a program which displayed a full screen of the same colour. When using that, the colours all came out much more saturated. I blame my TV card.

- Dave's and Bock's had smaller blocks of colour. That left rather small areas once the blurred edges had been cropped (blurring due to the low quality of the RF connection).

- If anyone else has a TV card, it'd be useful to compare.

- If I can sort out a composite video connection that should work much better.

Looking at the results, it looks like there is a little voltage "stealing" between R, G and B; for example, look along the row for B=0 and G=2. The green value should be constant but as the red value increases, the green decreases slightly. The same can be seen in most places.

If anyone would like the raw BMPs then just ask.

Many thanks again to the coders :)

Maxim

I missed out an important step...

1. Plug SMS into TV card instead of TV aerial
2. Change colour/contrast/brightness settings because you can't use the ones that are right for normal TV
3. Grab images

Because I only looked at the colour test programs I never noticed that the colours were off a bit.

So... I'll do it again. This time with a TV too.

Maxim

Good stuff... if you look at (e.g.) the last two line's green intensity:

203,203,197,190 200,199,193,198 (changes in Red affect Green: good info!!)

but:
255,255,255,255 255,255,255,255 (all the same?)

My guess is that the green is being maxxed out:
i.e. Real green value = about 0, 102, 205, 307, but getting clipped to 0,102,205,255. See what I mean?

Any chance of doing the test with the contrast turned down in order to keep the results within in the range of 0-255? That way we can see how the channels interact with each other at all levels and then adjust the contrast back up again later...

Quote

> Good stuff... if you look at (e.g.) the last two line's green intensity:

> 203,203,197,190 200,199,193,198 (changes in Red affect Green: good info!!)

> but:
> 255,255,255,255 255,255,255,255 (all the same?)

In all cases, when R, G or B is set to maximum, it came out as 255. I will try and tweak it so all 3 only just come to 255 or so; that, and I'll try and match it to a TV, maybe. Basically, when using the SMS, I have to turn the colour control to about 85% because the signal is not quite right, too powerful maybe, but I forgot to do it.

Anyway, I must give up for this evening... I have a driving test 8:30 Monday morning and I need to alter my sleep patterns to arrange to be awake for it...

Maxim

By sheer chance, when set to look right "by eye", most of the full-intensity blocks come to a value of 254 or so.

My apologies for the PNG, but GIF and JPG are too destructive (it's my own fault for having all that anti-aliased text).

FYI, I took a screengrab from Charles' 64-colour-at-once program, cropped out the blurred edges, then used PSP's Average blurring function over each block with the maximum aperture size, 3 times, giving an almost solid block of colour (if you look very closely at this image you will find variation of about +/- 1 in each channel in each block). Then I tidied it up, rearranged it to get the RGB separation shown, and added the values to the image (note that there may be transcription errors since I had to copy down all 192 numbers, if you find a discontinuity then check the actual colour yourself).

Maxim

Author	Message
Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Colour calibration..? Posted: Thu Jul 26, 2001 10:58 am
	I've noticed on some games recently that the colours are not quite right. In particular, Fantastic Dizzy uses a lot of browns which look much worse emulated than on-screen, and when playing through Micro Machines yesterday I kept thinking "that isn't right". So the obvious solution is... a homebrew program which will display large(ish) blocks of any colour on-screen (16 at a time would be useful and about the right size for each) which I will grab and attempt to smooth out to get "good" colour values. The problem is, I imagine the end result will still be quite variable. If I had a composite or RGB connection I expect the colours would be slightly different; then there's the issue of monitor gamma. My TV card does a pretty good job of converting between TV and monitor gammas though. Can anyone oblige with the code? I really need a book (or doc) to work from before I can even consider coding anything much myself. Maxim

Dave Guest	Posted: Thu Jul 26, 2001 10:48 pm
Dave Guest	Quote > So the obvious solution is... a homebrew program which will display large(ish) blocks of any colour on-screen (16 at a time would be useful and about the right size for each) Here's a program I made for you to do that. http://www.finalburn.com/colcheck.zip

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Results (57KB image) Posted: Fri Jul 27, 2001 10:37 am
	Many thanks to Dave, Charles and Bock for their various implementations. Here's that image: http://www.webtribe.net/m/maxim/temp/coltest1.png Taken from Charles' first program, mainly because his has the largest blocks of colour. It was all nicely aligned until I realised I had to rearrange it to get the RGB separation as shown, which is why it's a little misaligned. I hope the meaning of the numbers is clear. However, here are some things to consider: - My TV card is quite variable. Artificial solid blocks of colour do not come out as well as natural images. You'll probably have noticed the same if you ever tried to video your SMS, although the effect is much worse on a VCR. - Charles also wrote a program which displayed a full screen of the same colour. When using that, the colours all came out much more saturated. I blame my TV card. - Dave's and Bock's had smaller blocks of colour. That left rather small areas once the blurred edges had been cropped (blurring due to the low quality of the RF connection). - If anyone else has a TV card, it'd be useful to compare. - If I can sort out a composite video connection that should work much better. Looking at the results, it looks like there is a little voltage "stealing" between R, G and B; for example, look along the row for B=0 and G=2. The green value should be constant but as the red value increases, the green decreases slightly. The same can be seen in most places. If anyone would like the raw BMPs then just ask. Many thanks again to the coders :) Maxim

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Ignore that maybe Posted: Sat Jul 28, 2001 10:07 pm
	I missed out an important step... 1. Plug SMS into TV card instead of TV aerial 2. Change colour/contrast/brightness settings because you can't use the ones that are right for normal TV 3. Grab images Because I only looked at the colour test programs I never noticed that the colours were off a bit. So... I'll do it again. This time with a TV too. Maxim

Dave Guest	Maxxing out? Posted: Sat Jul 28, 2001 10:45 pm
Dave Guest	Good stuff... if you look at (e.g.) the last two line's green intensity: 203,203,197,190 200,199,193,198 (changes in Red affect Green: good info!!) but: 255,255,255,255 255,255,255,255 (all the same?) My guess is that the green is being maxxed out: i.e. Real green value = about 0, 102, 205, 307, but getting clipped to 0,102,205,255. See what I mean? Any chance of doing the test with the contrast turned down in order to keep the results within in the range of 0-255? That way we can see how the channels interact with each other at all levels and then adjust the contrast back up again later...

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Looks like it Posted: Sat Jul 28, 2001 11:11 pm
	Quote > Good stuff... if you look at (e.g.) the last two line's green intensity: > 203,203,197,190 200,199,193,198 (changes in Red affect Green: good info!!) > but: > 255,255,255,255 255,255,255,255 (all the same?) In all cases, when R, G or B is set to maximum, it came out as 255. I will try and tweak it so all 3 only just come to 255 or so; that, and I'll try and match it to a TV, maybe. Basically, when using the SMS, I have to turn the colour control to about 85% because the signal is not quite right, too powerful maybe, but I forgot to do it. Anyway, I must give up for this evening... I have a driving test 8:30 Monday morning and I need to alter my sleep patterns to arrange to be awake for it... Maxim

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Better results Posted: Sun Jul 29, 2001 3:00 pm
	By sheer chance, when set to look right "by eye", most of the full-intensity blocks come to a value of 254 or so. My apologies for the PNG, but GIF and JPG are too destructive (it's my own fault for having all that anti-aliased text). FYI, I took a screengrab from Charles' 64-colour-at-once program, cropped out the blurred edges, then used PSP's Average blurring function over each block with the maximum aperture size, 3 times, giving an almost solid block of colour (if you look very closely at this image you will find variation of about +/- 1 in each channel in each block). Then I tidied it up, rearranged it to get the RGB separation shown, and added the values to the image (note that there may be transcription errors since I had to copy down all 192 numbers, if you find a discontinuity then check the actual colour yourself). Maxim

Forums

View topic - Colour calibration..?