Hi, Rodrigo! Here again for your second critique!
Ok, so I've been seeing you post a lot of your graphics lately, mostly in the PYLC thread, and can I just say that I have been BLOWN AWAY by how much you've improved in such a short amount of time? Literally, everything I see nowadays is *stunning*. You've improved a hundredfold, and I've especially noticed it when it comes to text, composition, and colouring.
Still a huge fan of your current Joe Keery sig, but in addition to that I think my favourite thing in your gallery is +this Natty Ice banner—everything on this is perfect. The composition is great—you've done a great job with placing the different people and resizing to make the composition really interesting (I especially love how the smallest Natalia is facing away). The focal point is also so strong because of the way you've been mindful about which direction the models are directing their eyes. The colours, I can't even—the soft/spookly/unssatured undertones are illuminated brilliantly by the orange, and I love the way that colouring (perhaps you used a texture) also adds great shadows and depth to the Natalias. Finally, I love how you've been able to combine the use of textures and stock without overwhelming the graphic. Seriously, this is one of my favourite things that I have ever seen on TDA.
In addition to that brilliant banner, I also like +this blend, +this vertical sig, and +this CI; I think all of these are the best examples of really interesting/complex composition and great text. The text on the CI especially looks really natural; it's not too big, and I like how the blue colour fits nicely with the background but you're able to make it stand out with the pinkish stroke.
Ok, now I'll move on to addressing some of the things that you wanted me to look out for:
Making banners too stock heavy:
I actually don't see much of a problem with this! I think the Natalia Dyer banner I pointed out is a great example of an uncluttered stock banner, and as I already said, it's one of my favourite things in your entire gallery.
Going forward, I might even challenge you (if you want to get experimental that is—you don't have to if you don't think it fits your personal style) to use more stock on your graphics. Just like your placement of people and text can really make or break (as well as add complexity) to a graphic's composition, so can stock. I'd say don't worry if you feel like you're adding too much; it's hard to go overboard, but if you think that you're graphic is beginning to look cluttered, it's likely not because you've added too much stock, but rather because of how you're blending it. When working with stock (and as well as with people, of course), just make sure that your blending is neat, clean, and looks natural and you'll be golden.
Choosing text and placement:
Again, I think you've improved 100x with this. I mean, you managed to pull of using four different texts on +this blend and it looks great; that's something a lot of people wouldn't be able to do. Like last time, I'll just give a couple more suggestions of things to try out the next time you're experimenting with text. 
1.) Try working with some more scripty fonts! This is actually the opposite of what I end up telling most people; most people are too keen on throwing a cursive font on top of any graphic and forgo the san-serfic block-y fonts that you seem to use more often. Though these bold fonts look really great on a lot of your graphics (such as +here, for instance), there are a couple of your graphics where I think a cursive font would suit the mood better. Take +this signature; you've opted for a light background with light accessory colours, there are some flowers, etc.; on this particular signature, I think a cursive font rather than the simple san-serif one that you've chosen would cap off the serene, soft tone that I think you were going for.
2.) Make your text interact with the graphic with size and placement! Again, I think this is something you've already tried to tackle so props to you, but when you're making graphics, ask yourself, "How can my text placement and size contribute to the composition of the graphic?" +This banner I think is a great example of your text "interacting" with the rest of the graphic. I love how the word "riding"is kind of sloping down with Joe's shoulder, and I think that it's good that you chose this to be your main text while the remaining words are significantly smaller. Try a similar technique perhaps with +this signature. Right now, what should be the main text ("Melchior Gabor") isn't really interacting with the graphic or adding anything to it's composition. I'd try making one of these words (probably Melchior) larger than Gabor and then experimenting with how the shrinking or enlarging of "Melchior" (or just the "M" in Melchior) makes the graphic look.
- Another technique some people use to make their text "interact" with the rest of the graphic is seeing how it can wrap around one of their models, or how a model can partially hide some of their text. We can use the Melchior Gabor signature again (I don't think this suggestion would actually end up adding anything to your graphic, but we can just use it for the sake of example
). What would it look like if you changed the "M" in "Melchior" to a script font, made it significantly larger, and then put it near the guy's right shoulder (so our left)? You could then add a layer mask to make the the left-most art of the M hidden by the guy's shoulder so it looks like the first part goes behind the model but then peeks out in front of him later.
3.) Be mindful of text colour! One of the things I love about your graphics is how unafraid you seem to be to experiment with bold font choices and colours (I always say when it doubt, it's better to go bold than to go safe). However, make sure you're still keeping in mind the colour scheme of the rest of the graphic when you're making these decisions. On +this blend, for instance, I don't think the green colour of the font goes particularly well with the rest of the graphic. To me, it makes it stand out just a bit *too* much. However, if you prefer bold font colours, fear not because you can still make this happen. Considering the colour palette that you've gone with so far, what would a purple or even light blue (like the colour of the geometric shapes in the very bottom right hand corner) look like?
Ok, so to wrap things up I'm just going to focus on blending a little bit because it's just something I noticed. 
Blending
Ok, so it seems you've got a really strong command of hard cutting; on most of your graphics, the hard cuts look really good and clean. Make sure you strive for the same quality of blending when you're trying to do someone with, say, wispy hair. On +this signature, for instance, the edges of Natalia's hair are too soft so it doesn't look like she naturally fits in with that background. Probably the best advice I can give for blending wispy hair (as it's something that I still have to play around a lot with, too) is to not rely on the soft brush to get the job done, especially if the person has dark hair and you're trying to blend them into a light background, or vice versa. (A lot of the times you can get away with it though if you're blending a dark-haired person onto a dark background though). Something that I end up doing a lot of times that is really quick is that I'll take my model, paste them onto my canvas, and then either set them to darken/multiply (if dark hair on a light background) or lighten/screen (if light hair on a dark background). This will get rid of a majority of your background while still leaving a lot of detail in the hair. You'll then have to take a layer mask to get rid of the remaining parts of the background, but it mostly gets the job done. Then you'll have to duplicate that model layer, set them to normal, and a completely black layer mask, and then only fill in with white the parts of the model that still need additional detail (namely their face and body—if you don't do this, then there's a chance that part of your background will be peaking through their skin, which isn't good).
Ok, I think that's all I have to say for now! Here's a brief wrap-up of what I think are your strongest and weakest points:
Strong: COMPOSITION. Like literally, this seems to come so naturally to you. Colouring and hard cutting are also really good.
Points to work on: Experiment some more with text. Your text is already actually really good, but make sure you keep in mind those things I mentioned, primarily placement so it looks like it "fits in" with the rest of the graphic and colour. Also soft blending.
As always, feel free to contact me if you have any questions about anything I said and/or want me to elaborate on anything! Hope you've found this helpful, and I'll be looking for more of your graphics in the PYLC thread shortly!