In the context of the study of the integrated Sachs–Wolfe (ISW) effect, we construct a template of the projected density distribution up to redshift z ≃ 0.7 by using the luminous galaxies (LGs) from the Sloan Digital Sky Survey (SDSS) Data Release 8 (DR8). We use a photometric redshift catalogue trained with more than a hundred thousand galaxies from the Baryon Oscillation Spectroscopic Survey (BOSS) in the SDSS DR8 imaging area covering nearly one-quarter of the sky. We consider two different LG samples whose selection matches that of SDSS-III/BOSS: the low-redshift sample (LOWZ, z ∈ [0.15, 0.5]) and the constant mass sample (CMASS, z ∈ [0.4, 0.7]). When building the galaxy angular density templates we use the information from star density, survey footprint, seeing conditions, sky emission, dust extinction and airmass to explore the impact of these artefacts on each of the two LG samples. In agreement with previous studies, we find that the CMASS sample is particularly sensitive to Galactic stars, which dominate the contribution to the auto-angular power spectrum below ℓ = 7. Other potential systematics affect mostly the very low multipole range (ℓ ∈ [2, 7]), but leave fluctuations on smaller scales practically unchanged. The resulting angular power spectra in the multipole range ℓ ∈ [2, 100] for the LOWZ, CMASS and LOWZ+CMASS samples are compatible with linear Λ cold dark matter (ΛCDM) expectations and constant bias values of b = 1.98 ± 0.11, 2.08 ± 0.14 and 1.88 ± 0.11, respectively, with no traces of non-Gaussianity signatures, i.e. flocalNL=59±75 at 95 per cent confidence level for the full LOWZ+CMASS sample in the multipole range ℓ ∈ [4, 100]. After cross-correlating Wilkinson Microwave Anisotropy Probe 9-year data with the LOWZ+CMASS LG projected density field, the ISW signal is detected at the level of 1.62–1.69σ. While this result is in close agreement with theoretical expectations and predictions from realistic Monte Carlo simulations in the concordance ΛCDM model, it cannot rule out by itself an Einstein–de Sitter scenario, and has a moderately low signal compared to previous studies conducted on subsets of this LG sample. We discuss possible reasons for this apparent discrepancy, and point to uncertainties in the galaxy survey systematics as most likely sources of confusion.