link
Whilst not specifically about metal halide lamps, this clip demonstrates the dramatic effect colour temperature and spectrum response has on the colours in an image. This is why ceramic arc lamps look so much better than quartz arcs, even at lower wattages; not only is the arc shorter (allowing more light to pass through the triplet), but the spectrum response is closer to daylight (i.e. closer to the ideal black-body response).

A delightful little clip. Thanks for sharing it.

Colours on an object but Not on an Image!

Sorry, but I can't agree at all.

This Clip is exactly why people go off on tangents about colour quality of light. If you are building a projector that uses an LCD (all of us here !) then you need to use a light that was designed for LCD'.

The LCD passes only three colours and filters RGB only, that is ...it only passes tight range RGB. It has been designed and software colour corrected for Flourescent lights only. - note the prismatic spectrum of RGB in the Flourescent in the above video clip. Using the high quality output light from ceramics should actually reduce the throughtput of each Lumen of light because its not specifically designed to fit the RGB filters of the LCD. It becomes nothing but extra heat. The best light would be one that ONLY produces matched RGB and does not waste energy or effeciency on all the colours between. ie flourescent

Sodium would be good but requires extensive software correction to make up for the lack of one of the spectra. MH is better but does have a lot of energy wasted between the RGB bands and will still require correction for the spectrum imbalace. Remember that LCD's are designed for flourescents and ANY and ALL other light will be less effecient and require colour correction at the Software level. Unfortunatly, the only software correction is "Turning Down" the R,G or B until it matches the weekest contributor.

"Perfect colour" or high Colour index bulbs are for Museums and in my opinion are a waste of time for LCD projection .

Kirk

ahhh the light IS seen

short and concisely correct

ahhhhh, but not just any fluorescent will do ...... A Different PSOne Projector

See the difference?

Tgreenwood

Yes. Just as metal halide bulbs can have different colour temperatures, so can fluorescents.

BTW, has anyone ever found anythng definitive regarding the colour temperature of an LCD backlight?

Yes I know. Obviously the best one is the Backlight that came with the monitor right. Despite the fact it has a very poor colour rendering index because its not designed to light objects. It has the correct combination of metals balanced for the RGB filters. It has also then been hardware balanced for net colour throughput.

Other "off the shelf" Flourescents are again a waste of effort. Full spectrum just means it has a broad range of output. But you don't want broad range you only want RGB specific. AS a result your wasting all the lumens that fall between the bandpass spectral range. Yes it will give you better colour because it just happens to cotain a little more of the RGB your looking for but it still might only be 20% effecient for the job your doing.

I'm not suggesting using Flourescent bulbs for projection. if so you might as well just put a field lens against the mointor and go with what flys off the normal backlight. Instead you need projection specific bulbs. They are designed to only have spectral peaks where they are really needed. In lu of that, you need a very bright overkill source ( read MH) that contains atleast most of the RGB peaks and then you MUST have software colour correction.

with a klight meter its very easy to measure the colour level intensity needed. I'm sure its been done but I'll do it again for REf. Each panel will be different (Because they use different source of Backlights and different hardware adjustments) but I'll bet they are "generally" the same.

best,
Kirk

To stickgrip:

The color filters in the LCD have been proven to be wide band not narrow band.


"Full spectrum just means it has a broad range of output." That is a generic way of saying that the tube has a try-prosperous coating with a high color temp.

Many 5500K Flourescents will perform just like the backlights.

You said "Instead you need projection specific bulbs. They are designed to only have spectral peaks where they are really needed. In lu of that, you need a very bright overkill source ( read MH) that contains atleast most of the RGB peaks and then you MUST have software colour correction."


So what are these "projection specific bulbs"? If they are not "overkill" MH types then what are they made of?

I have never seen any PJ with any thing other than UHP which is a MH lamp or a xenon lamp which has the flattest wide band spectral-graph of any of the PJ lamps.

Both are high CRI lamps with low lumens per watt output.

In controlling and reproducing colour from the base RGB components you need to have tight control over the frequency and intensity of each base spectra. Only with tight Spectral lines are you able to control the desired effect of making a banana yellow out of a primary Red and Green. Given the very tight spectral lines provided by LCD florescent back light a more relaxed filter bandwidth can be selected and still not compromise the design. The filter in this case (LCD’s) only needs to make sure the other two frequencies don’t get through. Wider pass bands can therefore be employed because the next light producing region is a long way over.

The data I have for one filter set shows a fairly good control of Green and Blue but is relaxed in the High Red and IR range. This is probably because there is no need to filter off the top edge of the nearly invisible red light. It also will allow passage of any IR that would otherwise just go into heat in the panel. The result after you combine the filter set with the Back light output is you ultimately end up with those extremely tight frequencies spikes.
Click to view attachment

I guess the Point I am trying to make is that there is a synergy between the LCD colour filter , the Light source and the Gamma Correction Algorithm used in a LCD computer Monitor.

By stripping out the Back light and replacing it with some other source you will need to adjust either the Filter band pass and/or the gamma correction in order to get back to the colors as they were intended. Careful selection of a bulb will help the problem but finding one with close to the same output as the original Back light will be difficult.

Using a “Full” spectrum Light and “Loose” filters will result in a large reduction in colour reproduction and vibrancy. Tightening the filters will help but then a lot of that broad spectrum light will just get absorbed by the filters. Colour filters are essentially “transparent” at one frequency so it makes sense to only produce light in those ranges where it can pass through.

When using a projector lamp it is important to remember that the spectrum that is required for say film movies and overhead projectors is full spectrum. The more
“natural” the light the better the on screen rendition will be because we are trying to show the colour of the transparency which is itself is just a full spectrum filter.

Comparing this to discreet RGB filtered projection you need a light source and/or colour filter that results in a tight output of each RGB component so that the separate intensities can closely be controlled and “fool” the eye into seeing what is not really different colours at all but just a stimulation of three different cell receptors in the correct combination.

Any projector that uses RGB technology will use a select mix of ingredients in the bulb chemistry to make sure they have good useable spectral spikes that are close to the filter colours (and eye sensitivity range). These might be called broad spectrum or full spectrum because the range of spikes covers the RGB range. The projector manufacturer will then correct any shortcomings by software Gamma correction and end up with the final calibrated product.

In our DIY butchery we are guaranteed to upset the original precarious balance and will require a bit of wizardry to try and keep some semblance of intended colour quality.

Using full spectrum High CR index bulbs will simply result in more light energy being stopped by the filters and undesirably wider bands of RGB. By using colour correction charts and software adjustments we can likely use a wider range of bulbs and still get good representation.

Here is the colour filter that was used to produce the above output.
Click to view attachment
Its only intent is to keep the already tight bands in their proper slots and from leaking out the wrong pixel hole.

Cheers,
Kirk.

been saying this for ages too...

"I guess the Point I am trying to make is that there is a synergy between the LCD colour filter , the Light source and the Gamma Correction Algorithm used in a LCD computer Monitor.

Using a “Full” spectrum Light and “Loose” filters will result in a large reduction in colour reproduction and vibrancy. Tightening the filters will help but then a lot of that broad spectrum light will just get absorbed by the filters. Colour filters are essentially “transparent” at one frequency so it makes sense to only produce light in those ranges where it can pass through. "


I have never seen a florescent light that was full spectrum it is mainly a sales jargon.
Many of the Barco LCD PJ's that run HMI lamps? There fairly wide spectrum.



"When using a projector lamp it is important to remember that the spectrum that is required for say film movies and overhead projectors is full spectrum. The more
“natural” the light the better the on screen rendition will be because we are trying to show the colour of the transparency which is itself is just a full spectrum filter."


The main reason for flat wide spectrum lamps in film is mainly do to the total pain in the** of making color adjustments to an analog system. Any lamp with a perfect three spike RGB output would work fine if it had the same color temp that the film was mastered for. The HBO lamps that they use only have a slightly different spectrum output than an HMI lamp.

They use DC for a more stable arc and xenon for the smaller arc. They may be more color stable over time too.


"Comparing this to discreet RGB filtered projection you need a light source and/or colour filter that results in a tight output of each RGB component so that the separate intensities can closely be controlled and “fool” the eye into seeing what is not really different colours at all but just a stimulation of three different cell receptors in the correct combination."

The eye only has RGB color receptors so the soures only need to make RGB its not fooling the eye it the way we see color.

I found this

"Pigments are also found in cone cells. There are three types of cone cells, each of which contains a visual pigment. These pigments are called the red, blue or green visual pigment. The cone cells detect the primary colors, and the brain mixes these colors in seemingly infinitely variable proportions so that we can perceive a wide range of colors. Prolonged exposure to colors, for example when staring at a particular object, can cause fatigue in cone cells. This results in a change in the way that you perceive the color of the object that you are viewing."



"Using full spectrum High CR index bulbs will simply result in more light energy being stopped by the filters and undesirably wider bands of RGB. By using colour correction charts and software adjustments we can likely use a wider range of bulbs and still get good representation."
[/quote]


You make it seam like the PJ makers use some kind of magic lamps. The big problem with most MH lamps and many florescent lamps is they make little 610nm red. The green and blue are often fine with most but the red is often orange. This poops up the whole image.

The answer for most lamps is to increase the pressure and use a slightly different mix. A by product of making more red is they often make a flatter spectrum. We all want a perfect lamp with three spikes at 445nm, 525nm, and 6(edit)650nm but none exist.

The ceramic is special only because it has long life and makes some real 650nm red.

I can't agree with this. sodium puts out a very narrow band of yellow light. This is far different than a light source that puts out a wide band of light but looks yellow due to how our eyes interpret the mix. There is no amount of mixing for a sodium lightbulb that will show you green or blue. It just is not in the spectrum to pass through. You can't even mix what is in the sodium light to trick your eyes into seeing other colors because there simply is no other colors (besides yellow) to mix. However, if you have a wide spectrum source that simply looks yellow, but has other colors in the spectrum, then you can color correct by controlling the throughput through differnt filters.

Your right of course. I over simplified to make a point ant that was " sodium would be good (if only it contained the colours) but would require extensive correction" meaning you would have to turn the yellow off by reducing the leakage to the Red and green spectrum.
Here is a picture of a sodium bulb spectrum from the web.
Click to view attachment
Note that there is a good range of RGB available but the net value is likely only 10% of the total lumen output from the bulb. Filter bandpass would likely catch most of the light. Again the point being the more light that is availale ONLY at the filter range will give the best throughput.

Kirk

No, THIS is the spectrum from a sodium bulb:


A MASSIVE spike in the red region, no green, and a tiny amount of blue.

There seems to be some confusion about sodium lamps.
There are two types of sodium lamps. High pressure sodium (HPS) and low pressure sodium (LPS). LPS emits light of one wavelength (yellow). HPS emits light predominantly of yellow as well as small amounts of other bands.

DJ

Do you think that this lamp would reproduce colors as intended?


Spectrometer plot of a Philips 5000°K 13W CFL


Tgreenwood

I call true red about 650nm vary few lamps make that color as a primary spike. The CFL is still way sort here too and is why they only have a CRI of 82.

A look at a MEI short atc lamp.

Click to view attachment

This lamp make more 650nm red than any CFL lamp so it will have better reds on the screen too.

All the HMI /HMR/ MSR/MHI types of lamps also have shorter arcs' than the ceramic lamps. They are just vary inefficent.