Legoaizer - Help

Tutorial for Image Analysis

Version 5.0 only

Legoaizer has an effective approach to colour tuning. In many cases the selected colour is just not right for one's taste. That is due to the nature of matching colours: it's done by a (smart) algorithm, that tries to compare original and replacement colour, and selects one based on mathematical formulae. In many cases this is very well done, but in some cases... it's just not right.

For those cases Legoaizer has a so-called image analysis tool. This tool will analyse the source image for colours, and range of similar colours. It will then propose the best colour fit from the available colour palette. So far this is the 'traditional' approach. But now the user can select any colour from the original image, and make a different selection of colour. To further support colour customisation, also some sophisticated pre-processing is possible, where colours can be tuned and improved.

In this tutorial we will explain the basis of image analysis, show some examples and hopefully provide a solid starting point for the users to use this new tool. It is assumed that the user is familiar with all the bells and whistles of the interface (see here for more details).

Lesson 1: make an initial analysis   

The first step in image analysis is to prepare the basic parameters. The following parameters have influence on the image analysis:

These (relevant) settings in the main interface won't affect the image analysis:

Finally, the image analysis doesn't work either for the Mosaizer XV engine with actual pictures. That approach is completely different from the analysis approach.
The first step is to prepare the basic parameters: set the size, the colour table and the pre-filters. Then open the analysis window from the menu bar > 'Special functions' > 'Source analysis'. The interface is almost empty, so simply press the 'Analyse source file' button, and wait. The time of waiting will depend on the size of the mosaic. Usually this is done within a second. The picture below shows the result.
Now the difference between selected and original source colours is quite visible. For instance colour #8 (RGB 160-216-72) is quite different from the selected one: Dark tan. We will now show you how to select a different, hopefully better matching colour.

Lesson 2: select and replace colours    

Click on the name of the colour you wish to replace. In this case 'Dark tan'. You will now see a selection box with all the available palette colours. From here you can now select your replacement colour. But first, we like to draw your attention to the selection area in the picture on the left: some kind of 'moving' lines have become visible (because 'Show selected colour' is active). These are all the pixels that have colour #8. As you can see, just a few colours have this deviating colour. Not a big deal? Let's replace this with 'Bright green'. Select 'Bright green' from the list and you will notice a slight change in the picture on the right: the colour change is applied immediately (because 'Live update' is active).

 

This was a very simple lesson. To erase the change, press the 'Anlyse source file' button once again. The default selections are now shown again. To get rid of the moving lines, simply click anywhere in the image on the left.

Lesson 3: select a different colour match engine     

This is the fun stuff. The result in lesson 2 was based on the same colour match engine of the 'normal' mosaic creation. We will now select a different colour matching engine. For instance we will use the 'CIE L*a*b* ' engine. The image on the right is now different, and the table on the right now shows a different selection of colours. Still from the same palette, but just showing a different selection. Most importantly: it depends very much on the kind of image and your personal taste which colours you prefer. There is no objective truth here. It's your choice, not that of Legoaizer. It simply proposes the best fit using the selected colour engine (and colour space). Try a different colour matching engine, and see the effect of it. We have done this and all results are summarized in the picture below.

 

We have prepared a second example: skin colours. Here you can see how the different engines respond to skin colours.

Lesson 4: apply a dithering effect prior to colour matching      

Now comes the real fun stuff: pre-processing the colours of the original image, and then use a colour matching engine. A short word on dithering (as used here): we disperse pixels in such a way that with fewer colours we can still create the impression of a rich palette of colours. It's like cheating: up close you see points of individual colours, but from a short distance your brain interprets this blend of pixels as a solid colour.

The best approach is to select 'Palette colour' dithering. Here the dithering is done by only a choice from the current brick colour palette. This way colour matching and done during the dithering stage, and the colour matching engine is basically circumvented. The result is absolutely stunning (by the way, this is the same dithering as the 'normal' approach in older versions of Legoaizer, prior to version 5.0). This approach works best with 1x1 bricks only (or beads).

 

A second approach is to select the dithering combination of 'Original colour' with 'Median cut' and 'Error diffusion'. The result is less realistic, but more suitable for multiple bricks sizes. In the example below the HSL colour engine seems to create the best colour match.

The other dithering options are best suited for a different kind of mosaic: using almost a mono colour approach. We demonstrate this with a colour picture, but we want to use a bluish colour scheme to create an arty mosaic. Here we show 5 examples of the effective use of colour dithering, with a colour source image.

Example 1: no dithering. Nice fill of bluish colours, and great for 'posterizing' effect, but this not what we had in mind.

Example 2: black and white dithering. Only two 'colours' are required. It selects the best fit with the black.

Example 3: error diffusion method in black white. Same as example 2, but not more pixelated, and more depth of the mosaic result.

Example 4: fixed halftone and ordered. The colours are quite well translated in 8 shades of gray (into blue, we had only 14 shades of blue in our palette). We added some contrast to the original: a value of 8.

Example 5: same as example 4 (adding contrast), but now we used fixed halftone, where 64 shades of gray are created, and a 4x4 dithered pattern. Sharp, sufficient colour depth and a very good impression of the original. 

Lesson 5: making a mosaic with after image analysis       

When you are finally satisfied with your colours, press the button 'Create mosaic' in the analysis windows (not the usual button!). It will circumvent the traditional way of creating a mosaic, and use the current right image and the selected colours instead. The output is as usual: a (brick) mosaic image and a blueprint excel spreadsheet. Ready for building. The close-up image below shows the results from example 5.

What can you do now?       

Just keep playing with settings of dithering, colour matching engines, and colour spaces. For almost every kind of image there will be an ideal combination of setting. It's your 'quest' to find it. It's hard to provide more guidelines. Hopefully, the examples above are motivating to simply do it.


Copyright © 2015-2018 APP Helmond