05-23-2018 10:18 AM
We have 1 bit unsigned image (only one color lines, texts, symbbols... ) and we want to ppublish in apollo core. We need "1bit output with tranparent backgroundd", only one color in the output. What is the (best) way (what format, how to prepare it, how to ppublish)? I found jpeg2 lossless compression, it looks well in the "base scale" but when we use zoom out - it looks any "midcolor" pixels between black lines and background. lossless ECW provide nearly different colors in the background. IMG - I'm not able to set transpoarent background.... Many thanks for help.
05-24-2018 03:37 AM
How is your original data?
If it is 1bit data, you can already use it in APOLLO Core, you just need to configure the styling properly.
For instance, you can use a formula to mask all pixels with the "0" value:
if INPUT1 = 0 then null else 1
And then, use a look up table.
FYI, you can create you own look up table (C:\Program Files\Hexagon\ERDAS APOLLO\lib\etc\erm\lut and restart IIS + Core Console)
05-24-2018 10:22 AM
Hi Florent, many thanks for your response. My original data is tif file that is recoded in the imagine. What is the best format for the best performance in the apollo?
(I did the styling for the IMG that you wrote, and it looks for all scales; is it the 1bit IMG the best format? or is it there any way how to convert it into the JP2/ECW eithout lose quality for smaller scales ?)
05-29-2018 05:19 AM - edited 05-29-2018 05:19 AM
Difficult to say without testing which format is the best. But with ECW you cannot have 1bit images (maybe the same with JP2 I don't know). So the ECW may be bigger than the IMG and it would need some work to convert the IMG into ECW 8bit without loosing quality (and 0 should be masked with the alpha channel).
I would say it is maybe easier to test if the performances are already acceptable with the IMG because the data is already there.
07-18-2018 06:00 PM
I think the best way to determine the best format for your 1bit images is to do some benchmarking.
To do this I would create some sample data in a number of different formats then test the performance on Apollo using Apache JMeter. Make sure the sample data is larger than the available memory on the machine so you can test reading the image from storage instead of memory cache.
The test plan you create will need to hit the WMS getmap request with a large selection of random coordinates.
The reason I suggest testing with JMeter is because you cannot test performance as a single user connected to a WMS service in a client. This is because you cannot push the limits of the service because you will probably only need 15 tiles per second for it to feel quick, this will not scale to multiple users possibly requiring hundreds of tiles per second. With JMeter you will be able to determine if your server can support multiple users.
Since the images are 1bit I have a feeling that IMG will meet your performance requirements but you will not know for sure without testing. You may find that ECW or JP2 is better, especially if you can use compression on your data.