That Targetseq = Nimblegene is our (limited!) experience. We tried both and found Targetseq outperformed (just), with slightly higher on target reads, and there a buffer that facilitates the hybridisation process when one includes adaptors/barcodes prior to the capture (in addition to barcode "blocking" oligos). What that buffer is I don't know (I would like to!)
I gather from Agilent they are developing something similar, their kit promises multiplexing of 8 - 16 exomes prior to capture. They (agilent) tell me that ontarget reads remain high. Any experience?