Open source spectrophotometer?

[Richard Hughes]

Hey all,

I've been thinking recently about also creating an open source display
spectrophotometer. This would be in addition to the low-price ColorHug
colorimeter. It'd be a lot more expensive (at least £200) but of
course this depends on the quantity that I build and the
specification. I'm basically wondering if anyone would be interested.
I'd probably start prototyping around December time, just to see if I
can get something basic working.

Richard.

[Allan Savolainen]

On Fri, 25 May 2012 19:13:41 +0300, Richard Hughes <hugh...@gmail.com>
wrote:

Atleast I would be interested on such a thing.

- Allan

[Peter Bex]

I would be very interested in a spectrophotometer,

Peter

[Grant Goodyear]

On Fri, May 25, 2012 at 11:13 AM, Richard Hughes <hugh...@gmail.com> wrote:

I've been thinking recently about also creating an open source display
spectrophotometer.

I'd be interested (and I'd buy one just to support the project), but I'd be much more interested in one that could read printer patches as well.   Is the design that much different between the two?

-Grant-
--
Grant Goodyear       
web: http://www.grantgoodyear.org   
e-mail: gr...@grantgoodyear.org

[Josenivaldo Benito Junior]

Count +1

--
Josenivaldo Benito Jr.
PU2LBD

*Por Aurélio Buarque de Hollanda,  elite, do francês élite, significa “o que há de melhor em uma sociedade, minoria prestigiada, constituída pelos indivíduos mais aptos”.

[Richard Hughes]

On 25 May 2012 17:19, Grant Goodyear <gr...@grantgoodyear.org> wrote:
> I'd be interested (and I'd buy one just to support the project), but I'd be
> much more interested in one that could read printer patches as well.   Is
> the design that much different between the two?

Then the cost goes up even higher, as it'd need to provide a
wide-spectrum light source, and it's kinda tricky to find a low cost,
low power LED to do that. Advice welcome.

Richard.

[Ferry Huberts]

On 25-05-12 18:19, Grant Goodyear wrote:
> On Fri, May 25, 2012 at 11:13 AM, Richard Hughes <hugh...@gmail.com
> <mailto:hugh...@gmail.com>> wrote:
>
> I've been thinking recently about also creating an open source display
> spectrophotometer.
>
>
> I'd be interested (and I'd buy one just to support the project), but I'd
> be much more interested in one that could read printer patches as
> well. Is the design that much different between the two?

me too, I'd buy one for sure!


--
Ferry Huberts

[Kurt Roeckx]

A LED itself only transmits 1 wavelength. But some of the white
LEDs use phosphor to covert it a wide spectrum like a
fluorescent-lamp. See
https://en.wikipedia.org/wiki/Light-emitting_diode#Phosphor-based_LEDs


Kurt

[Stijn Hoop]

Cautious +1 although it depends a lot on the price for me. A lot more
expensive than ColorHug is of course expected given the rather more
complicated functionality, but I would compare the price to non-free
alternatives as well. I'm willing to pay a bit more for the open source
feature, but as said it depends on how much I'd have to pay for that.

--Stijn

[Fernando Michelotti]

I'm really interested in one.

[Chris Lilley]

On Friday, May 25, 2012, 6:13:41 PM, Richard wrote:

RH> Hey all,

RH> I've been thinking recently about also creating an open source display
RH> spectrophotometer.

Display only, i.e. only works with emissive sources not reflective ones?

That would seem to limit the usefulness (and sales) since the first thing people seem to want a spectro for is to calibrate a printer.

Spectros can calibrate a screen of course and on the plus side are much less sensitive to the precise type of backlight technology; on the downside they tend to have more problems measuring dark colours.

As an example, my i1D3 gives better display profiles than my ColorMunki Photo, despite being around half the price.

If the primary intent is to get better quality screen profiling (at higher cost than colorhug) then a colorimeter whose spectral responses are closer to a linear combination of the XYZ functions, and with optics to gather more light too increase accuracy in dark areas, might be a better goal.

In other words, if the intent is an open source competitor to the i1D3, while the colorhug is an open source competitor to the Huey or the Spyder II.

If on the other hand the primary motivator is the lack of an open source spectro, then I believe it needs to be capable of reflective as well as emissive measurements so that it can profile printers.

Being able to sample the ambient light (through some sort of diffusing hemisphere) would also be a useful capability, for the photographer crowd (report color temperature in Kelvin, color rendering index, deltaE from the blackbody and daylight loci etc. The code for which is in Argyll spotread.

(On a colormunki, spotread -a -s -S).

RH> This would be in addition to the low-price ColorHug
RH> colorimeter. It'd be a lot more expensive (at least £200) but of
RH> course this depends on the quantity that I build and the
RH> specification. I'm basically wondering if anyone would be interested.
RH> I'd probably start prototyping around December time, just to see if I
RH> can get something basic working.

I would be interested but the level of interest also depends on the capabilities. And the cost, and the accuracy (as I already have a ColorMunki).

I note that the ColorMunki has a fixed, non-removable UV cut filter, the i1Pro comes in two otherwise indentical models with or without a fixed UV cut, and the i1Pro 2 has a switchable UV cut filter (and a switchable incandescent or D50 light source).

Having a removable UV cut filter (or a holder for one, even if the filter is not supplied) would be a real plus.

Since I can imagine Richard wincing and saying 'cost' after each of the above paragraphs, screw-in ambient diffusers or uv-cut filers could be extra-cost options or could already be available from third parties, it then simply being a case of making the spectro compatible so they can be fitted.

--
Chris Lilley Technical Director, Interaction Domain
W3C Graphics Activity Lead, Fonts Activity Lead
Co-Chair, W3C Hypertext CG
Member, CSS, WebFonts, SVG Working Groups

[Jan Köster]

Me too.

[Richard Hughes]

On 25 May 2012 20:43, Chris Lilley <ch...@w3.org> wrote:
> Display only, i.e. only works with emissive sources not reflective ones?

Well, good question. I was thinking of starting the design with
emissive, and leaving PCB space and power budget for a wide gamut LED.

> That would seem to limit the usefulness (and sales) since the first thing people seem to want a spectro for is to calibrate a printer.

Right, agree.

> Spectros can calibrate a screen of course and on the plus side are much less sensitive to the precise type of backlight technology; on the downside they tend to have more problems measuring dark colours.

Yes, I was thinking of including a ColorHug-style colorimeter built
into the spectro, so it can generate it's own CCMX matrices.

> Being able to sample the ambient light (through some sort of diffusing hemisphere) would also be a useful capability

Agreed, although it makes the package much harder -- I have to be
careful as a rotating design like the (original) ColorMunki is very
patented.

> I would be interested but the level of interest also depends on the capabilities. And the cost, and the accuracy (as I already have a ColorMunki).

Right. A calibration tile (plural?) would also be a nice feature too.

> I note that the ColorMunki has a fixed, non-removable UV cut filter, the i1Pro comes in two otherwise indentical models with or without a fixed UV cut, and the i1Pro 2 has a switchable UV cut filter (and a switchable incandescent or D50 light source).

Yes, it just makes the package harder to design. Adding a removable UV
filter is only a couple of quid extra.

> Having a removable UV cut filter (or a holder for one, even if the filter is not supplied) would be a real plus.

Right.

> Since I can imagine Richard wincing and saying 'cost' after each of the above paragraphs

Hah, you read my mind :)

> , screw-in ambient diffusers or uv-cut filers could be extra-cost options or could already be available from third parties, it then simply being a case of making the spectro compatible so they can be fitted.

Well, the filters themselves are not that expensive, but the metal
collets all have to be designed to hold them. I'm going to think about
the goals and possible designs for the new few months and I'll post
photos of progress along the way :)

Thanks for all your help,

Richard.

[James Cloos]

>>>>> "RH" == Richard Hughes <hugh...@gmail.com> writes:

RH> Then the cost goes up even higher, as it'd need to provide a
RH> wide-spectrum light source, and it's kinda tricky to find a low cost,
RH> low power LED to do that. Advice welcome.

Everything I see uses tungsten halogen. 5W seems to be the right fit.
I doubt any LED would do better than 5W in any case. Long life (5K to
20K hours) 5W xenons are inexpensive and as close to black-body as we
have, I suspect. And a set of leds covering near ir to violet would
require just as much cooling capacity.

You might need a 12V power supply though. Or perhaps the lighting
assembly could have its own power supply, separate from the meter's.
(Such a separation is not unsusual, from what I can tell.)

-JimC
--
James Cloos <cl...@jhcloos.com> OpenPGP: 1024D/ED7DAEA6

[Matthias Urlichs]

On Fri, 2012-05-25 at 14:49 -0300, Fernando Michelotti wrote:
> I'm really interested in one.

Me too; I also second the requirement for printer profiling (using a
small oncandescent bulb as the light source, not a 'white' LED, if at
all possible).


--
-- Matthias Urlichs

[Thilo Fromm]

Hello Richard,

If it supports printer profiling I'd buy one for sure, too.

A hardware consideration: I recommend using an Atmel ATMega controller
the next time 'round, and have it run Arduino. Way better Linux support
(and MacOS, and Windows...), way less hassle and boiler plating. They're
available in amazingly small package sizes, too.

Regards,
Thilo

[Pascal de Bruijn]

On Sun, May 27, 2012 at 4:45 PM, Thilo Fromm
<thilo.a...@googlemail.com> wrote:
>
> If it supports printer profiling I'd buy one for sure, too.
>
> A hardware consideration: I recommend using an Atmel ATMega controller the
> next time 'round, and have it run Arduino. Way better Linux support (and
> MacOS, and Windows...), way less hassle and boiler plating. They're
> available in amazingly small package sizes, too.

Back when the ColorHug development started, there was no Arduino which
was capable of exposing an HID USB interface (at least not easily).

Possibly something like this might be worth considering as a base...
http://www.sparkfun.com/products/11098

However, there are always concerns with experience, changing platform
will probably slow development down considerably.

Regards,
Pascal de Bruijn

[Richard Hughes]

On 27 May 2012 15:57, Pascal de Bruijn <pmjde...@pcode.nl> wrote:
> However, there are always concerns with experience, changing platform
> will probably slow development down considerably.

Right; agreed. I'm also not sure 16MHz is going to be fast enough. I
really want to use something like a 3648 pixel linear CDD array, and
that's going to need a ~2Mhz data bus -- 1/8 isn't a lot of headroom
from a timing point of view. I was thinking either a PIC32 or even a
small cheap FPGA, although the number of people knowing VHDL is going
to be pretty small.

Richard.

[Thilo Fromm]

Hello Pascal,

>> If it supports printer profiling I'd buy one for sure, too.
>>
>> A hardware consideration: I recommend using an Atmel ATMega controller the
>> next time 'round, and have it run Arduino. Way better Linux support (and
>> MacOS, and Windows...), way less hassle and boiler plating. They're
>> available in amazingly small package sizes, too.
>
> Back when the ColorHug development started, there was no Arduino which
> was capable of exposing an HID USB interface (at least not easily).

Sorry, I didn't mean to be reproachful. This was pure forward-thinking.

> Possibly something like this might be worth considering as a base...
> http://www.sparkfun.com/products/11098

You don't even have to start with existing hardware as long as you use
one of the ATMegas supported by Arduino. I was more thinking about
development support and active open source communities.

> However, there are always concerns with experience, changing platform
> will probably slow development down considerably.

Not so much. It's highly embedded C in both cases, no MMU, that's it.
Some details do differ, sure, that's what documentatin and datasheets
are for. And Arduino is rather well-documented. Trust me, I know both
worlds, I'm with the LoLShield
firmware,(http://code.google.com/p/lolshield/) and now I'm here :)

With Arduino you would win a large, diverse, and active community for
(potential) firmware hacking, though, and a wide selection of tools for
all major OSes plus less boiler plating, since there are a lot of
ready-to-use libraries. I bet some of the guys would start messing with
the firmware just because the device can run Arduino.

Regards,
Thilo

[Thilo Fromm]

Hello Richard,

>> However, there are always concerns with experience, changing platform
>> will probably slow development down considerably.
>
> Right; agreed. I'm also not sure 16MHz is going to be fast enough. I
> really want to use something like a 3648 pixel linear CDD array, and
> that's going to need a ~2Mhz data bus -- 1/8 isn't a lot of headroom
> from a timing point of view. I was thinking either a PIC32 or even a
> small cheap FPGA, although the number of people knowing VHDL is going
> to be pretty small.

Hey, this would be an opportunity to re-activate knowledge I didn't use
for about 10 years :)

OTOH if you need so much processing power why not go for a full-blown
low-power embedded Linux system? With a state-of-the-art ARM CPU you can
easily get 16 bit wide static memory interfaces at ~100MHz or more for
sampling. It's tough to design and more expensive to produce, though.

Regards,
Thilo

[Alexander Wright]

On Sunday 27 May 2012 17:21:53 Thilo Fromm wrote:
> OTOH if you need so much processing power why not go for a full-blown
> low-power embedded Linux system? With a state-of-the-art ARM CPU you can
> easily get 16 bit wide static memory interfaces at ~100MHz or more for
> sampling. It's tough to design and more expensive to produce, though.

Isn't this one of the uses a Raspberry Pi would be ideal for?

Are any of these interfaces suitable?

http://elinux.org/Rpi_Low-level_peripherals


Alexander.

[Pascal de Bruijn]

That's probably overkill and overcomplicated...

Richard is probably talking using something akin to this:

http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,892,893&Prod=CHIPKIT-UNO32

Maybe loosely basing off a future Arduino Due might be an option:

"Arduino Due, a major breakthrough for Arduino because we’re launching
an Arduino board with a 32bit Cortex-M3 ARM processor on it. We’re
using the SAM3U processor from ATMEL running at 96MHz with 256Kb of
Flash, 50Kb of Sram, 5 SPI buses, 2 I2C interfaces, 5 UARTS, 16 Analog
Inputs at 12Bit resolution and much more. "

In the end, the largest factor will probably be what Richard likes to
work with, as he's doing 99% of the actual work :D

Regards,
Pascal de Bruijn

[Richard Hughes]

On 27 May 2012 18:16, Pascal de Bruijn <pmjde...@pcode.nl> wrote:
> In the end, the largest factor will probably be what Richard likes to
> work with, as he's doing 99% of the actual work :D

Agreed :)

Richard.

[Graeme Gill]

Chris Lilley wrote:

> I note that the ColorMunki has a fixed, non-removable UV cut filter, the i1Pro comes in two
> otherwise indentical models with or without a fixed UV cut, and the i1Pro 2 has a switchable UV
> cut filter (and a switchable incandescent or D50 light source).

It's not that the ColorMunki has a filter, it has two limitations: The White LED
doesn't emit any UV, and the diffraction grating efficiency drops quite
steeply in the blue region. I'm not sure what the i1Pro2 actually does yet, although
I hope to have a better idea soon.

Graeme Gill.

[Graeme Gill]

Thilo Fromm wrote:
> A hardware consideration: I recommend using an Atmel ATMega controller the next time 'round, and
> have it run Arduino. Way better Linux support (and MacOS, and Windows...), way less hassle and
> boiler plating. They're available in amazingly small package sizes, too.

Yes :- I've been using the Arduino for a couple of projects, and the
compile environment was pretty easy to get going, even outside
the Arduino IDE. Seems like a nicer instruction set (ie. the
compiler doesn't have to be as clever), and maybe the USB
would work better ?

Graeme Gill.

[Graeme Gill]

Richard Hughes wrote:

> small cheap FPGA, although the number of people knowing VHDL is going
> to be pretty small.

Verilog is a lot closer to "C", and probably more appropriate for
smaller projects. (I found Alter's AHDL to be much easier to
get going on that either Verilog or VHDL, but it is proprietary.)

Graeme Gill.

[Matthias Urlichs]

On Sun, 2012-05-27 at 16:57 +0200, Pascal de Bruijn wrote:
>
> Back when the ColorHug development started, there was no Arduino which
> was capable of exposing an HID USB interface (at least not easily).
>

Things have improved since then. There's a couple of nice libraries for
the current crop of USB-capable Atmega chips which you can use to teach
the things a rather large number of tricks. ;-)

--
-- Matthias Urlichs

[Chris Lilley]

On Monday, May 28, 2012, 2:29:18 AM, Graeme wrote:

GG> Chris Lilley wrote:

>> I note that the ColorMunki has a fixed, non-removable UV cut filter, the i1Pro comes in two
>> otherwise indentical models with or without a fixed UV cut, and the i1Pro 2 has a switchable UV
>> cut filter (and a switchable incandescent or D50 light source).

GG> It's not that the ColorMunki has a filter,

Thanks for the clarification. In fact, since the X-Rite pages on the ColorMunki Design & Photo are mainly marketingese and rather shy on hard technical details, my source for that statement was the Argyll supported instruments list:

"Unlike the Eye-One Pro, the ColorMunki is only available in a U.V. Cut (Ultra Violet filtered) model. "
http://www.argyllcms.com/doc/instruments.html#ColorMunki

GG> it has two limitations: The White LED
GG> doesn't emit any UV, and the diffraction grating efficiency drops quite
GG> steeply in the blue region.

Aha, I see. It would be good to add that info to the supported instruments page, and perhaps modify the existing statement to something like

"Thus, unlike the Eye-One Pro 1, the ColorMunki is effectively only available in a U.V. Cut (Ultra Violet filtered) model."

GG> I'm not sure what the i1Pro2 actually does yet, although
GG> I hope to have a better idea soon.

My understanding is that the i1Pro1 has a tungsten filament incandescent light source (so basically Illuminant A - which would therefore have little violet or blue energy and, I would have thought, very little ultraviolet). Given what you say above, I guess that the diffraction grating efficiency covers a wider range of wavelengths so that it is at least able to measure what little light is reflected back at those wavelengths.

I'm also aware that the i1Pro1 (and the ColorMunki spectro, and likely other spectros as well) give best results at higher light intensities and worst at lower ones; so they have problems measuring dark colors.

If that is all correct, then it would seem that the accuracy and precision in the near-uv, violet and blue region would be better, on the i1pro1, for emissive sources than for reflective ones.

I gather that the i1Pro2 has a dual illuminant - a tungsten filament incandescent plus a UV LED and that its 'dual scan' mode takes two sets of measurements, one with each illuminant, then computes what the results would have been if the illuminant had been D50.

See for example
http://www.luminous-landscape.com/reviews/accessories/eye_one_pro_ii.shtml

I came across an interesting industrial spectro where the spectro itself and the illuminant are entirely different devices, and both connect to the reading head via optical fibres. One of their illuminators used four different LEDs (a warm white, cool white, cyan, and blue/uv) to approximate a relatively even spectrum.

They also had an optical cable with two central 'read' fibres surrounded by illuminant fibres. One read fibre was a uv/vis and the other a vir/nir and they went to two different spectros configured to cover different wavelength ranges.

[Graeme Gill]

Chris Lilley wrote:

> "Unlike the Eye-One Pro, the ColorMunki is only available in a U.V. Cut (Ultra Violet filtered)
> model. " http://www.argyllcms.com/doc/instruments.html#ColorMunki

Hi,

The statement is in brackets because it's a nomenclature translation, but I'll see if
I can emphasise the non literal nature a bit more.

> My understanding is that the i1Pro1 has a tungsten filament incandescent light source (so
> basically Illuminant A - which would therefore have little violet or blue energy and, I would
> have thought, very little ultraviolet). Given what you say above, I guess that the diffraction
> grating efficiency covers a wider range of wavelengths so that it is at least able to measure
> what little light is reflected back at those wavelengths.

The i1pro2 seems very similar to the i1pro in fundamentals. The illuminant and optics
appear unchanged, apart from the addition of two supplemental UV LEDS. While incandescent
illumination has relatively low UV output, this is magnified in an instrument setting
because the reflected light level is divided by the illuminant level. The result is
that a quite noticeable level of fluorescence can be excited.

> I'm also aware that the i1Pro1 (and the ColorMunki spectro, and likely other spectros as well)
> give best results at higher light intensities and worst at lower ones; so they have problems
> measuring dark colors.

Every instrument does. The i1d3 is significantly better because it collects light
from a much larger area than usual with these low end instruments.

> If that is all correct, then it would seem that the accuracy and precision in the near-uv,
> violet and blue region would be better, on the i1pro1, for emissive sources than for reflective
> ones.

Perhaps, but the diffraction grating and possibly fibre optics seems to limit it.

> I gather that the i1Pro2 has a dual illuminant - a tungsten filament incandescent plus a UV LED
> and that its 'dual scan' mode takes two sets of measurements, one with each illuminant, then
> computes what the results would have been if the illuminant had been D50.

Yes. It's a similar idea to Argyll's FWA compensation, and using a UV
source that you can switch on and off was something I suggested in
the related paper. Nice to see that X-Rite have taken the suggestion up :-)
My initial experiments were of a very similar nature - comparing UV cut
to UV included, and 'A' to simulated D65 illiuminant using the Spectrolino.

> I came across an interesting industrial spectro where the spectro itself and the illuminant are
> entirely different devices, and both connect to the reading head via optical fibres. One of
> their illuminators used four different LEDs (a warm white, cool white, cyan, and blue/uv) to
> approximate a relatively even spectrum.

There are now some viewing booth lights that use a whole set of LEDs to allow you
to dial the spectrum up. I think X-Rite may be doing a similar thing in one
of their high end spectros.

Graeme Gill.

[Josenivaldo Benito Junior]

How about an ARM Cortex M3 (Stellaris or NXP would be my choices)?

You can compile with GCC with no pain. Can you FreeRTOS to make things even portable and simple... sounds more openSource than a FPGA. 

Just food for thoughts 

--
Josenivaldo Benito Jr.
PU2LBD

*Por Aurélio Buarque de Hollanda,  elite, do francês élite, significa “o que há de melhor em uma sociedade, minoria prestigiada, constituída pelos indivíduos mais aptos”.

[Robert]

I would be interested in this product.  Speed and low-light accuracy always important :)

I'm in.

On Friday, May 25, 2012 9:13:41 AM UTC-7, Richard Hughes wrote:

Hey all,

I've been thinking recently about also creating an open source display
spectrophotometer. This would be in addition to the low-price ColorHug
colorimeter. It'd be a lot more expensive (at least £200) but of
course this depends on the quantity that I build and the
specification. I'm basically wondering if anyone would be interested.
I'd probably start prototyping around December time, just to see if I
can get something basic working.

Richard.

[Richard Hughes]

On 21 July 2012 20:01, Visigoth <tino...@gmail.com> wrote:
> P.S. maybe it would be wise to share the design of the PCB etc. of this
> future spectrophotometer with the community before putting it into actual
> production, since it appears to be a quite more complex device than the
> ColorHug.

Sure, I thought a lot about this for the ColorHug. I don't think a lot
of people understand high-speed PCB design, but certainly a bill of
materials would let people help with things like "That LED isn't very
wide gamut, this one is better".

Richard.

[vnor...@gmail.com]

I would go for it too!!!

In the meantime, I will go for a ColorHug :)

[Graeme Gill]

vnor...@gmail.com wrote:
> In the meantime, I will go for a ColorHug :)

If you want an economical and accurate spectrometer
right now, then it's hard to go past the ColorMunki
at US$400 == �255 (although I guess the current $50 rebate
may not apply outside US/Canada).

Graeme Gill.

[JLD]

My two cents is really a highly biased perspective: it would be tremendous to have an open-source spectrophotometer that can do print profiles. My use-case is to provide an open-source orientation for technology solutions in developing economies. It's not just cost, it's about the knowledge participation. Secondly, I really enjoy photography and find that creating print output can be very limiting without proper paper/printer profiles. Artisan papers, and media variation (e.g., cloth, plastic transparencies, etc.) make personal photography printing a unique experience. While not a hugely expanding segment of enthusiasts, personal, hi-quality printing is rising. Creating paper/printer profiles on an open-source platform would be terrific.

Colorhug is tremendously valuable, would a spectrophotometer (Spect-o-hug) meet or exceed the capabilities of closed-source options ?  

Jack

[jim hatch]

Yes definitely.  Mainly for calibration of home theater, but would like it for printer photography also.

On Wed, Feb 13, 2013 at 9:10 PM, Louis <lph...@lebureau.ca> wrote:

Yes I'm in too!

--
You received this message because you are subscribed to the Google Groups "colorhug-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to colorhug-user...@googlegroups.com.
Visit this group at http://groups.google.com/group/colorhug-users?hl=en-GB.
For more options, visit https://groups.google.com/groups/opt_out.