Greek - στ is very tight

[EbenSorkin]

Greek - στ (sigma tau) is very tight

It looks like this is because τ (tau) is itself spaced very tightly with zero on the left and 50 on the right like lowercase latin t. However the tau is optically more alanogous to Cyrillic te ( which is space 50 on both sides ) so I suggest taking an approach to the left side bearing which is looser or perhaps even a mirror value to the right side.

Similarly, it is important to be able to easily pick out or identify the right stem of sigma and I think it should also be looser.

image from Roboto flex wght 100

[davelab6]

Please could you confirm the exact design space location you are seeing this on? And which file exactly, and from where :)

[EbenSorkin]

The font I am using comes from here: https://github.com/googlefonts/roboto-flex/blob/main/fonts/RobotoFlex%5BGRAD%2CXOPQ%2CXTRA%2CYOPQ%2CYTAS%2CYTDE%2CYTFI%2CYTLC%2CYTUC%2Copsz%2Cslnt%2Cwdth%2Cwght%5D.ttf

I am having trouble with the newest release of Octopus so I am using the TN tool until that's resolved.

Here are screenshots using the updated font and that tool.

1. wght 100, wdth 100, opsz sizes are matched to set sizes and are 144, 42, 24, 16, 12, slnt 0 GRAD 0

In this sample the gap I expect to see between ττ is present at but I would argue is slightly too small in 42, 24, 16, 12, but is missing from 144.

Swapping GRAD to 150 improves things very very slightly but not meaningfully so.

Italic set to full -10 opens things very slightly for all opsz except again 144

[EbenSorkin]

looking at 300 wght and italic & roman - similar observations

[EbenSorkin]

The size 12 text was wider than it should have been in the last two samples here is corrected with a doubled t for reference

and the light 100 wght

[EbenSorkin]

wght 500

and slnt -10

[EbenSorkin]

At 700 wght the gap I think would be desirable is visible with tau at 144 but is less substantial than I think is desirable at the smaller sizes. The gap I'd like to see between sigma and tau flickers between the opsz.

There is a little variation in the effect in slnt-10 but it is very similar

[EbenSorkin]

I will ask Kostas if he prefers to see the outstroke of sigma ( σ )match the thickness of the crossbar in t and tau.

[EbenSorkin]

900 roman shows the gaps present again at 144 between doubled tau but still not between sigma and tau

In italic the pattern seem to match roman