Okay, nevermind. I skimmed through his post over at wherever that is, and I think he's not using the best of angles with some shots for scaling purposes.
There will be errors of course but not big ones. Maybe it would be good if I repeat and explain my scaling procedure once again.
Trade Federation battleship
First I will start with what I think is a most accurate scaling picture:
+
http://www.geocities.com/idesdjurdja/tf ... _width.jpg
Here we see a top down view of the battleship from which we can derive the length ratio between the "arm" of the battleship and it's radius with almost no perspective error. The radius is 947px while the width of the arm is 358px. This means that the ships radius is about 2.645 times greater than the width of the arm or that the width of the ship is 5.29 times greater than the width of the arm.
From this picture:
+
http://www.geocities.com/idesdjurdja/arm_width.jpg
we can get a fairly accurate hangar to arm length ratio. According to my measurements the hangar is 90px wide while the arm is 203px wide. This means that the arm is about 2.256 times greater than the width of the hangar. There could be a slight perspective error in this picture but even eyeballing the picture it's clear that the arm is something over two times wider than the main hangar.
Since the arm is 2.256 times wider than the hangar and 5.29 times shorter than the the ship, the ship is 11.93 times wider than the hangar.
Even if we assume that the landing craft is exactly as wide as hangar(which it is not) then 3170m diameter for the battleship yields a 265m width for the landing craft. Alternatively 370m width for the landing craft yields a 4414m width of the battleship. There is no way that both 370m figure for the landing craft and 3170m for the battleship can be correct at the same time.
Next examine the following picture:
+
http://www.geocities.com/idesdjurdja/hangar_width.jpg
I measured the width of the landing craft as being 361px while the width of the hangar is 386px. This means that the hangar is 1.069 times wider than the landing craft.
Thus the battleship is 12.76 times wider than the landing craft.
Once again there is a perspective error but it only makes the landing craft look that it's wider in relation to the hangar than it really is.
I think it is safe to conclude that the battleship is 12.76 times wider than the landing craft which means that at least one but likely both figures from the ICS are wrong.
Landing craft
Observe the following picture:
+
http://www.geocities.com/idesdjurdja/la ... width2.jpg
I measured the heigth of the landing craft as being 71px while the half-width of the craft is 235px. This means that the total width of the craft is 6.62 times greater than it's heigth. The heigth of the landing craft is 84.46 times smaller than the width of the battleship. Any perspective error from this picture will result in underestimation of the width to heigth ratio and consequently the width of the ship.
From the following picture:
+
http://www.geocities.com/idesdjurdja/la ... heigth.jpg
we can determine landing craft heigth to STAP heigth ratio. The stap is off in the distance and can not be easily recognized especially on this JPEG picture but anyone with a TPM DVD can easily watch the part of the film in question and see that there really is a STAP flying towards the craft.
Anyway the heigth of the ship is 326px while the heigth of the STAP is 12px. So the STAP is 27.16 times shorter than the landing craft.
We can now easily determine that the STAP's heigth is 2294.5 times smaller than the battleship's width.
Finally we have this picture:
+
http://www.geocities.com/idesdjurdja/st ... thedit.jpg
If the battledroid is 1.8m tall than the STAP(actally STAP+battledroid's head and torso) is 2.57m tall.
This means tha the battleship is 5,897m wide.
Conclusion
For a STAP's heigth of 2-2.57m we get a width of the Trade Federation battleship being between 4589m and 5897m.