Several scientific topics linked to the observation of extended structures around astrophysical sources (dust torus around AGN, disks around young stars, envelopes around AGBs) require imaging capability with milli-arcsecond spatial resolution. The current VLTI instruments, AMBER and MIDI, will provide in the coming months the
required high angular resolution, yet without actual imaging. As a rule of thumb, the image quality accessible with an optical interferometer is directly related to the number of telescopes used simultaneously: the more the apertures, the better and the faster the reconstruction of the image. We propose an instrument concept to
achieve interferometric combination of N telescopes (4 ≤ N ≤ 8) thanks to planar optics technology: 4 x 8-m telescopes in the short term and/or 8 x 1.8-m telescopes in the long term. The foreseen image reconstruction quality in the visible and/or in the near infrared will be equivalent to the one achieved with millimeter radio interferometers. Achievable spatial resolution will be better than the one foreseen with ALMA. This instrument would be able to acquire routinely 1 mas resolution images. A 13 to 20 magnitude sensitivity in spectral ranges from 0.6 to 2.5 μm is expected depending on the choice of the phase referencing guide source. High dynamic range, even on faint objects, is achievable thanks to the high accuracy provided by integrated optics
for visibility amplitude and phase measurements. Based on recent validations of integrated optics presented here an imaging instrument concept can be proposed. The results obtained using the VLTI facilities give a demonstration of the potential of the proposed technique.