Hyperconjugation bond stretching from

Evidently, the C-H bond should hyperconjugate more strongly in acetaldehyde than it does in propene, and the C-H bond length difference as we go from 0° to 90° in torsion angle should increase more. And that s what we find. The calculations show that in this case the C-H bond length increase in acetaldehyde is 0.0047 A, vs. 0.0030 A in propene. This particular type of bond stretching has also been referred to in the literature as the carbonyl effect ... 

This is a kind of torsion-stretch cross term but different from the one where the central bond changes with torsion angle. There has been some considerable debate about the existence and origin of the hyperconjugative effects, but low-temperature X-ray crystallographic experiments on appropriate compounds together with ab initio calculations certainly reveal a detectable effect. 

It is instructive to think of hyperconjugation in terms of the valence bond structures, and then relate those structures to whatever motions (stretching, bending, torsion combinations) that are qualitatively predicted by those structures. And the forms of the equations that will describe those elements are usually evident from the valence bond... 

---