[MCN-L] Color to Grayscale (mpara...@gallery.ca)
I just want to thank Will Real for having the courage to say what I've been advocating for years: scan it in RGB, storage be damned. Time and again, we see this gradual realization. While there is little that is "cheap" when it comes to digitizing any form of cultural heritage materials, storage is one area where the costs have continuously been decreasing. If a user truly needs a grayscale image for whatever purposes, it is possible to automate the conversion from RGB to grayscale so that it can be done "on demand." This has been possible since the beginning of the web, through the use of ImageMagick and other scriptable tools. But you can never get back the color information if you never had to begin with or chose not to keep it. Best to all, Dennis ~~ If your first move is brilliant, you?re in trouble. You don?t really know how to follow it; you?re frightened of ruining it. So, to make a mess is a good beginning. ? Brian Eno On May 21, 2013, at 8:57, Will Real wrote: > Marianne, Kevin, et al., > > I've been following this discussion with interest. > > I was involved in a large B+W negative scanning project starting back in > 2003, and we debated this issue extensively. Back then, the cost of storage > was a significant factor. The images were to be 6000 pixels on the long > side, stored as 16-bit TIFFs in both unedited and edited versions, for each > of 80,000 negatives in the collection. > > We decide to convert to grayscale not only because of the storage concerns, > but also because we concluded that it was most important to capture the > relative "density" values of the negative, the information on the negative > that enables the image to be rendered in positive form. For this purpose > grayscale is a reasonable analog for the actual density values in the > negative. The primary aim of the project was access to the images in > positive form, rather than capturing detailed characteristics of the > negatives themselves as physical objects. We ended up scanning the > negatives in RGB, and then converting to Grayscale in Photoshop, usually > with the Green channel, which generally had the best tonal characteristics > and the least scanner noise compared with the other color channels. > > Some of the negatives had pink, blue, or yellow stains, originating in the > anti-halation layer of the negative. These negatives posed a further > dilemma. We did not want to lose the condition information because it might > be useful for future researchers. But we also did not want to memorialize > the staining as part of the positive image itself. In the end we decided > that in these cases we would convert to grayscale using the color channel > that mimimized the stain the most. This gave us a grayscale image that was > truest to the image represented by the negative. We also saved a > lower-resolution version of the scan in RGB, as a reference for the > condition aspects of the negative as a physical object. > > There was a group of about 500 B+W prints in the collection. All of the > prints were scanned in RGB in order to preserve the color characteristics > of the prints. Even though the prints were all monochrome, very few were > strictly grayscale images. > > If I were starting a project like this today, the storage issue would not > drive the decision at all. I think I would save the masters in RGB, and > when converting the images into positive form, I would probably only > convert to grayscale when the staining in the negatives would compromise > the accuracy of the positive image. > > Will > > PS for what it's worth, we also included with each negative scan a density > step-wedge reference. With these references it is possible to extrapolate > the actual densities present in the negative, rather than simply having the > relative densities represented as grayscale values.We considered storing > actual densitometer measurements with the metadata, but the economics and > time constraints of the project did not allow it. So the step-wedges > offered a way to capture at least some useful density information much more > expediently. > > > > On Sat, May 18, 2013 at 8:44 AM, Kevin Sprague wrote: > >> Marianne >> >> The simple answer to your question is that no - if you have a color image >> and you reduce it to grayscale you lose data - its that simple. Scanning or >> photographing in color is going to provide a greater range of options for >> data retrieval from an image over time than a greyscale image. It you want >> to see evidence of this for yourself, in photoshop open a color image and >> go to window>channels - you will see the RGB color challens (assuming that >> you are in RGB color space) displayed in the palette. You can click on each >> of the channels and the image should show you just that channel - usually >> in greyscal - and you can see how the different color channels manage data >> independently. For instance, in many photographs, the "luminance" data - >> that is,
[MCN-L] Color to Grayscale (mpara...@gallery.ca)
Marianne, Kevin, et al., I've been following this discussion with interest. I was involved in a large B+W negative scanning project starting back in 2003, and we debated this issue extensively. Back then, the cost of storage was a significant factor. The images were to be 6000 pixels on the long side, stored as 16-bit TIFFs in both unedited and edited versions, for each of 80,000 negatives in the collection. We decide to convert to grayscale not only because of the storage concerns, but also because we concluded that it was most important to capture the relative "density" values of the negative, the information on the negative that enables the image to be rendered in positive form. For this purpose grayscale is a reasonable analog for the actual density values in the negative. The primary aim of the project was access to the images in positive form, rather than capturing detailed characteristics of the negatives themselves as physical objects. We ended up scanning the negatives in RGB, and then converting to Grayscale in Photoshop, usually with the Green channel, which generally had the best tonal characteristics and the least scanner noise compared with the other color channels. Some of the negatives had pink, blue, or yellow stains, originating in the anti-halation layer of the negative. These negatives posed a further dilemma. We did not want to lose the condition information because it might be useful for future researchers. But we also did not want to memorialize the staining as part of the positive image itself. In the end we decided that in these cases we would convert to grayscale using the color channel that mimimized the stain the most. This gave us a grayscale image that was truest to the image represented by the negative. We also saved a lower-resolution version of the scan in RGB, as a reference for the condition aspects of the negative as a physical object. There was a group of about 500 B+W prints in the collection. All of the prints were scanned in RGB in order to preserve the color characteristics of the prints. Even though the prints were all monochrome, very few were strictly grayscale images. If I were starting a project like this today, the storage issue would not drive the decision at all. I think I would save the masters in RGB, and when converting the images into positive form, I would probably only convert to grayscale when the staining in the negatives would compromise the accuracy of the positive image. Will PS for what it's worth, we also included with each negative scan a density step-wedge reference. With these references it is possible to extrapolate the actual densities present in the negative, rather than simply having the relative densities represented as grayscale values.We considered storing actual densitometer measurements with the metadata, but the economics and time constraints of the project did not allow it. So the step-wedges offered a way to capture at least some useful density information much more expediently. On Sat, May 18, 2013 at 8:44 AM, Kevin Sprague wrote: > Marianne > > The simple answer to your question is that no - if you have a color image > and you reduce it to grayscale you lose data - its that simple. Scanning or > photographing in color is going to provide a greater range of options for > data retrieval from an image over time than a greyscale image. It you want > to see evidence of this for yourself, in photoshop open a color image and > go to window>channels - you will see the RGB color challens (assuming that > you are in RGB color space) displayed in the palette. You can click on each > of the channels and the image should show you just that channel - usually > in greyscal - and you can see how the different color channels manage data > independently. For instance, in many photographs, the "luminance" data - > that is, the gradient of greys between Balck and White - is often very well > represented in the Red channel. Often the blue channel will contain color > "noise" - the blue wavelength is difficult for modern sensors and scanners > to capture but there may be critical picture information in this channel. > You can play with combining channels and see what emerges. > > It your question regarding greyscale vs. color motivated by storage > concerns? A greyscale image will contain only about 1/3 the data of the > same RGB image and therefore takes up less storage space on hard drives. > This used to be a consideration for insititutions but with the declining > cost of storage and the availability of very large 2-3TB hard drives, the > economic argument of storing large, high resolution images is ceasing to be > an issue. At my business we maintain around 500,000 high res files on about > 10TB or storage and the overall cost is not much more than about $4,000 at > this point, and dropping every day. > > So - the simple answer is - scan and shoot in color, save in the largest > color range you can (i.e. 32bit vs 16bit), explore f
[MCN-L] Color to Grayscale (mpara...@gallery.ca)
Marianne The simple answer to your question is that no - if you have a color image and you reduce it to grayscale you lose data - its that simple. Scanning or photographing in color is going to provide a greater range of options for data retrieval from an image over time than a greyscale image. It you want to see evidence of this for yourself, in photoshop open a color image and go to window>channels - you will see the RGB color challens (assuming that you are in RGB color space) displayed in the palette. You can click on each of the channels and the image should show you just that channel - usually in greyscal - and you can see how the different color channels manage data independently. For instance, in many photographs, the "luminance" data - that is, the gradient of greys between Balck and White - is often very well represented in the Red channel. Often the blue channel will contain color "noise" - the blue wavelength is difficult for modern sensors and scanners to capture but there may be critical picture information in this channel. You can play with combining channels and see what emerges. It your question regarding greyscale vs. color motivated by storage concerns? A greyscale image will contain only about 1/3 the data of the same RGB image and therefore takes up less storage space on hard drives. This used to be a consideration for insititutions but with the declining cost of storage and the availability of very large 2-3TB hard drives, the economic argument of storing large, high resolution images is ceasing to be an issue. At my business we maintain around 500,000 high res files on about 10TB or storage and the overall cost is not much more than about $4,000 at this point, and dropping every day. So - the simple answer is - scan and shoot in color, save in the largest color range you can (i.e. 32bit vs 16bit), explore file formats with minimal or lossless compression, and buy big hard drives! Kevin Sprague Studio Two www.studiotwo.com On Sat, May 18, 2013 at 8:00 AM, wrote: > Send mcn-l mailing list submissions to > mcn-l at mcn.edu > > To subscribe or unsubscribe via the World Wide Web, visit > http://mcn.edu/mailman/listinfo/mcn-l > or, via email, send a message with subject or body 'help' to > mcn-l-request at mcn.edu > > You can reach the person managing the list at > mcn-l-owner at mcn.edu > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of mcn-l digest..." > > > Today's Topics: > >1. Re: Color to Grayscale (MParadis at Gallery.ca) > > > -- > > Message: 1 > Date: Fri, 17 May 2013 16:08:21 + > From: > To: > Subject: Re: [MCN-L] Color to Grayscale > Message-ID: <9B3B404398037F4FA199183563FADE9C0D477AA1 at Medusa.NGC.MBAC> > Content-Type: text/plain; charset="iso-8859-1" > > Here's another approach; > > Open a copy of your original in full colour and simply de-saturate the > colour values in hue and saturation controls. Works extremely well in not > dropping all the grey subtleties. > > Good luck! > > MARK PARADIS > CHIEF, MULTIMEDIA SERVICES-CHEF DE SERVICES MULTIM?DIA > > NATIONAL GALLERY OF CANADA, MUS?E DES BEAUX-ARTS DU CANADA > 380 SUSSEX DRIVE, OTTAWA, ONTARIO K1N 9N4 > > PH. 613-990-1788, FAX. 613-991-2680, CELL?613-797-0558 > > > > -Original Message- > From: mcn-l-bounces at mcn.edu [mailto:mcn-l-bounces at mcn.edu] On Behalf Of > Frank E. Thomson > Sent: April-30-13 3:23 PM > To: Museum Computer Network Listserv > Subject: Re: [MCN-L] Color to Grayscale > > In PhotoShop you can convert to a gray scale in a manner that keeps > maximum tonal range. After saving you cannot convert back to color so that > information is lost. > > Frank Thomson > Asheville Art Museum > Mailing address: PO Box 1717, Asheville, NC 28802 Street address: 2 South > Pack Square, Asheville, NC 28801 > 828.253.3227 t > 828.257.4503 f > fthomson at ashevilleart.org > www.ashevilleart.org > > Our Vision: to transform lives through art > > -Original Message- > From: mcn-l-bounces at mcn.edu [mailto:mcn-l-bounces at mcn.edu] On Behalf Of > Marianne Weldon > Sent: Tuesday, April 30, 2013 7:42 AM > To: mcn-l at mcn.edu > Subject: [MCN-L] Color to Grayscale > > I've been led to believe that converting color images to grayscale > digitally does not loose information, but have no actual 'proof' of this. > Is anyone aware of any documentation or publications on this topic? > Additionally, I know many people that choose to scan black and white images > in color then convert to greyscale.again...any useful data or > discussions on this out there? > > Thank you! > > > Marianne Weldon > Fellow, The American Institute for Conservation Collections Manager of Art > and Artifacts > 202 Canaday > Bryn Mawr College > 101 North Merion Avenue > Bryn Mawr, PA 19010 > office 610-526-5022 > mweldon at brynmawr.edu > > See our collection online at: Triarte.br
