The ImageZebra Service ​

Using ImageZebra ​

  • Can I place multiple test targets in a single image to upload and analyze?
    Not at the moment. Currently we only support a single test target per image.

  • What targets are supported?
    See the list of supported test targets for all currently supported targets.

  • Can I use my own custom reference values?
    This feature is on the list of new upcoming features. We hope to announce this ability soon!

  • I want to use the M-19-21 specification. Do you support that?
    Yes! The FADGI 2023 specification calls this "FADGI 2023 Modern Textual Records". This is identical to the M-19-21 specification.

  • Is there a size limit to the uploaded images?
    ImageZebra supports input images up to about 250 Meg in either Tiff or Jpeg format. If your Tiff files are bigger than that, possibly utilizing a losless compression like LZW may help

  • Does ImageZebra support all GoldenThread targets?
    ImageZebra supports the ISA Golden Thread Device level target, as well as all three ISA Golden Thread Object level targets. (All three sizes.)

  • Can ImageZebra create an .icc profile from an upload?
    The generation of .icc profiles from an uploaded target is on our roadmap, but not currently available on the platform

  • Can ImageZebra correct an image?
    This is certainly on our roadmap, but is not currently available. We are considering two options: Building an .icc profile to encapsulate the correction, or actually applying the correction to customer images.

Help with understanding the results ​

  • Our image is able to meet three-four stars across all categories except for Lightness Uniformity (which is greyed out). Why is it greyed out?
    In the situation where a test target does not have the elements required to adequately measure a metric, the results are greyed out. The ImageZebra platform is capable of measuring a variety of test targets, however, not all targets have the elements required for all the metrics. The results panel always shows all the FADGI metrics regardless of whether the current test target is capable of measuring each metric. If you are using the Golden Thread Object Level target which is a thin long target, it does not cover a large enough area to adequately measure uniformity across the entire field of view. Therefore, the results show the Uniformity metric as greyed out, since it can not be measured with the test target you submitted. In order to evaluate Uniformity, the Golden Thread Device Level target would be recommended.

  • Our image is able to meet three-four stars across all categories, but the Low Noise Warning does not show any stars. Why?
    This means that the measured noise level is extremely low and does not achieve an expected level for most test patterns and imaging systems. This metric however is not a requirement for FADGI compliance. It is listed as a Warning only, which is a recommendation that you check to ensure the digital count values are not being clipped or processed in some non-linear manner, resulting in nearly constant values for a uniform patch. (Most test targets and imaging systems always have some small level of variation). Occasionally an image will be found that has variation/noise, generally in the range of 0.22 - 0.24 (standard deviation of L* values) as a minimum level. The low noise warning, according to the FADGI guidelines, is set to a threshold level of 0.25. Such an image is just slightly below the threshold, therefore, the 0-star performance will be reported. Since this is a "Warning only" (Not a FADGI compliance requirement) and the fact that the image is right at the threshold level, it should not be a concern.

Problems ​

  • I received a “Target could not be identified” error, but I think my image is OK. What causes this?
    If you are receiving a “Target could not be identified” error when uploading an image of a supported target, here are some things to keep in mind:

    • An image should only contain a single test target. See the list of supported test targets.
    • Make sure the entire test target is in the image. For GoldenThread object level, that includes the circular black registration dots at either end of the color patches.
    • The color patches should be at least 25 pixels in each dimension.
    • The image should be defined using 24 or 32 bits per pixel. 1 and 8 bit per pixel images are not supported.
    • The color or brightness of the image should not be too far off.
    • The image should not be excessively noisy.
    • The color patches need to be at least 1% of the maximum dimension of the image.
      For example, say the image has dimensions 4800 x 6400 pixels. Since the maximum dimension is 6400, the color patches need to have a size of at least 1% of that, or 64 pixels.

  • Occasionally a user may attempt to submit an image that contains a large area of content (such as artwork) and a small test target at the edge. This type of image might fail the last rule - the 1% rule. An easy way to fix this is to simply crop out the test target portion of the image, and resubmit that cropped region. This makes the color patches much bigger with respect to the size of the image

  • I received an error saying “Target could not be identified. This may be because the image is low resolution (M ✕ N pixels)”
    The image was not high enough resolution. The patches in the test target need to be at least 25 ✕ 25 pixels. That means the test targets themselves need to be at least the following size:

    1. Golden Thread Device Level: 800 ✕ 1,200 pixels
    2. Golden Thread Object Level: 1,000 ✕ 100 pixels
    3. ColorChecker SG: 1,500 ✕ 1,000 pixels